phData cert progress β source of truth
This file is the canonical progress record read by /cert-study (cron + on-demand).
Update it after any study session, exam attempt, or blocker change.
The skill reads it to derive verdicts; it never overwrites entries β only appends.
Partner-portal access (Cert 2 gate)
| Status | Date confirmed | Notes |
|---|---|---|
| π‘ PATH IDENTIFIED β likely self-serve | 2026-07-13 | Colleague who passed (902) confirms: create account at anthropic-partners.skilljar.com using phData email β register for the exam via Pearson VUE (mandatory since July 1). Registration instructions: anthropic-partners.skilljar.com/claude-certified-architect-foundations-certification/486716. Founder action: create the Skilljar account (~15 min) β gate either clears or we learn the real blocker. |
| β not yet provisioned (pre-2026-07-13 status) | β | Required before Cert 2 (Anthropic) study materials are accessible. Action: follow up with phData once onboarding settles. |
2026-06-13 (iMessage): Founder is about to start a Claude Architect cert prep course with practice exams β so either the partner-portal gate above has effectively cleared or he's using external/public prep. β Confirm gate status on next
/cert-study. Advised diagnostic-first: take ONE practice exam cold before the course (pretesting effect + triages which modules to skim vs sit through β high baseline since he builds agent architectures daily), and save remaining exams for spaced post-course retrieval + a near-final readiness check.
Cert 1 β Snowflake GenAI Specialty (GES-C02)
Deadline: 2026-11-22 (CORRECTED 2026-08-24 by /cert-study β escalators are order-agnostic per founder's 2026-08-17 correction; Anthropic CCA-F passed first and banked the 3-month leg, so Snowflake is now the 6-month leg. The "90d from job start / 2026-08-24" framing below and in [[milestones]] is the pre-correction label and is stale β do not cite it.)
Deadline (stale, pre-correction): 2026-08-24 (90d from job start 2026-05-26)
Exam scheduled: β not yet scheduled
Track selected: β (Track A = 7-10 day sprint recommended; Track B = 3-4 week steady)
Exam registration: not yet registered
Study log
| Date | Day/Topic | Time spent | Notes |
|---|---|---|---|
| no entries yet |
Weak areas (populated by /cert-study quiz)
none logged
Attempt log
| Date | Outcome | Score | Notes |
|---|---|---|---|
| no attempts |
Cert 2 β Anthropic Claude Certified Architect β Foundations
Deadline: 2026-11-22 (180d from job start 2026-05-26) Exam scheduled: β not yet scheduled (gated on partner-portal access) Current phase: Phase 1 β D1 Agentic Architecture (target: Jun 2026)
Study log
| Date | Phase/Domain | Time spent | Notes |
|---|---|---|---|
| formal materials gated on partner-portal access | |||
| 2026-06-10 | D1+D4+D3 quiz (via /cert-study on-demand) | ~15 min | 4/5 β see quiz log. Weak area: D3 API specifics (Message Batches, tool_choice, streaming) β daily Claude Code use doesn't teach the raw API surface. |
| 2026-08-17 | Cram day β founder booked 1-5p ET focus block, asked for weak-area map | β | Ray sent the ROI-ranked list from the exam-2 miss analysis (test-craft β hook scoping β API surface β act-vs-ask β context economy) + out-of-scope don't-study list. Offered mid-block quiz. NB: cramming ~3 months ahead of the 11-22 deadline β possibly targeting the ~08-26 first-3-months escalator window; exam-scheduled status still unconfirmed. |
| 2026-08-17 | PASSED β Claude Certified Architect β Foundations, 850/1000 (bar 720) (real exam, same day as the cram) | β | Founder texted the Credly congratulations screen 17:40 ET, score 850 confirmed 17:50 ("comfortably passed," his calibration). Badge issuer Anthropic, accept-link via Credly. Ray advice: accept with issued email (bwilson@phdata.io, Skilljar-registered), then add bwilson668@gmail.com secondaryβprimary in Credly for portability. Banks the 3-month escalator leg (~08-26 cap) 9 days early. FOUNDER CORRECTION same day: escalators are order-agnostic β the "Snowflake by 08-24" framing Ray had been carrying was WRONG; Snowflake GenAI is the 6-month leg, "another 3 months," ~late Nov 2026. OPEN: verify partner-portal gate row can be closed Β· re-pace Snowflake plan to the real window. Cram-day protocol (miss-data ROI list + out-of-scope cuts) validated β reuse for Snowflake. |
| 2026-08-17 | ESCALATOR #1 PAID β cert submitted to phData same evening, +$5k locked, base now $200k | β | Founder 20:18 ET: "officially locked in my first tranche of salary boost. I'm at $200k base now." Compounds into 401k contributions, match, and bonus qualification. Escalator-paperwork item CLOSED (he did it himself, no draft needed). Remaining escalator: Snowflake GenAI (~late Nov) β $205k. |
Weak areas (populated by /cert-study quiz)
- D3 API specifics (logged 2026-06-10, quiz 1): Message Batches API β founder "no clue" (taught: 50% discount, β€24h no-SLA, single-shot/no tool loop). Drill next:
tool_choicemodes (auto/any/forced), streaming event sequence (message_start β content_block_delta β message_stop), extended-thinking token budgeting. Root cause: daily Claude Code use never touches the raw API surface. CONFIRMED on exam 2 (Q7 tool_choice, Q33 prefilling, Q52 MCP notifications) β promote to active drill. - Hook scoping (logged 2026-06-27, exam 2; recurring from exam-1 Cluster A): tool-specific hooks only fire for that tool β Write hook β Bash
echo > file; comprehensive protection needs Write+Bash hooks (+.claudeignore+OS perms); exit-code 0=allow/non-zero=block. Drill from [[2026-06-14-cca-harness-mechanics-cheatsheet]]. - "Most-complete answer" test-craft (logged 2026-06-27, exam 2; ~6 Qs): on BEST/most-robust/MOST-addresses stems, pick the layered/both-halves/one-notch-further option, not the clean single-mechanism one. Not a knowledge gap β a test-taking habit. Highest-ROI fix.
- Act-vs-ask calibration (logged 2026-06-27, exam 2; ~4 Qs, judgment): the autonomy-threshold rubric (additiveβaccommodate; material changeβsurface; divergent outputsβpresent-both-if-cheap; self-param-change w/o goal-checkβrisk).
- Context economy (logged 2026-06-27, exam 2; ~2 Qs): right-size output to consumer's decision; progressive summarization across sessions.
Attempt log
| Date | Outcome | Score | Notes |
|---|---|---|---|
| 2026-06-14 | PASS (practice, pre-study) | 50/60 = 83% (pass = 72%) | First practice exam, COLD, before the course. Instructor benchmark: most get 50-60% after the course β founder well above post-course cohort pre-study. 10 misses (founder read: only 2 real guesses, 1 broken question). Pattern + fixes β [[2026-06-14-cca-harness-mechanics-cheatsheet]]. |
| 2026-06-27 | practice (borderline) | 42/60 = 70% | Pass line recorded as 72% (β οΈ source unverified β CONFIRM real cut score; founder reports 70% passes). Finished 12 min to spare. Down 13 pts from exam 1 β but it's targeting not regression: misses concentrate in 1 test-craft pattern + 2 knowledge gaps + 1 judgment axis. All 18 misses analyzed β [[2026-06-27-cca-practice-exam-2-miss-analysis]]. |
Miss pattern (practice exam 1, 2026-06-14)
- Cluster A β harness mechanics/timing (~half, highest ROI): Stop-hook per-turn (Q48 β), PostToolUse timing (Q10 β exam answer shaky vs current docs: PostToolUse CAN modify output before context), MCP
resources/updatednotifyβclient-refetch (Q27 β), subagents parent-mediated (Q34 β), skill triggers intent-not-keyword (Q42 β). Verified vs docs.claude.com via claude-code-guide. - Cluster B β plausible-neighbor definitions: idempotencyβ atomicity (Q12), episodic/semantic by specificity-not-duration (Q22).
- Cluster C β "BEST/canonical" vs pragmatic-builder answer: fan-out/barrier/fan-in (Q24), subagent-delegation for context isolation (Q35).
- Low-signal near-duplicate-answer traps: Q26 (has an editing artifact in its own explanation), Q42.
- D4: clean sweep (0 wrong).
Miss pattern (practice exam 2, 2026-06-27)
All 18 misses β full breakdown [[2026-06-27-cca-practice-exam-2-miss-analysis]]. Five threads, ROI-ranked:
- "Most-complete answer" under-selection (test-craft, ~6 Qs, highest ROI): Q4/Q5/Q19/Q56/Q18 (+Q23 flavor). On BEST/most-robust/MOST-addresses stems, founder picks a clean single-mechanism answer; key is the layered/both-halves/one-notch-further option. Fix = a test-taking rule, zero study.
- Hook scoping (knowledge, ~3 Qs, RECURRING): Q42/Q5/Q23. Tool-specific hooks only intercept that tool β a Write hook does NOT catch Bash
echo > filewrites; exit-code 0=allow/non-zero=block. Was Cluster A on exam 1 β persisting. Re-read [[2026-06-14-cca-harness-mechanics-cheatsheet]]. - API surface (knowledge, ~3 Qs): Q7/Q33/Q52.
tool_choiceforces a tool call (+downstream validation); prefilling{locks JSON vs override framing; MCP capability-change notifications are protocol-level/transport-agnostic. = the "D3 API specifics" gap flagged 2026-06-10 (tool_choice named then), now confirmed. - Act-vs-ask calibration (judgment, ~4 Qs): Q8/Q15/Q17/Q55. Miscalibrated autonomy threshold, errs both ways. Rubric: additive discovery β accommodate+note; material scope/resource change β surface options; divergent outputs unspecified β present both if cheap else ask; self-modifying params w/o goal-check β anti-pattern.
- Context economy (knowledge-judgment, ~2 Qs): Q46/Q9. Match output granularity to consumer's decision (summary not 500 rows; progressive summarization across sessions).
Exam guide update (2026-07-02)
Founder shared updated official exam guide (v0.2, June 30 2026) post-maintenance-window. Domain weights/structure unchanged vs. 05-27 baseline. New: full exam mechanics now confirmed (60 Q, 120 min, $125 fee, 12-month validity, 720/1000 pass bar confirmed official), 27 granular task statements, 6 named scenarios, 12 worked sample Qs w/ explanations, and an explicit Out-of-Scope list (drops streaming API/SSE, computer use, vision, fine-tuning, MCP hosting/infra, pricing/rate-limit calc, auth/billing, tokenization). Action: the "D3 API specifics" weak-area label above is mislabeled β tool_choice/Message Batches are D4 (Prompt Eng & Structured Output) under current numbering, not D3. Drop the queued streaming-event-sequence drill (now confirmed out-of-scope). Full diff β [[2026-07-02-cca-exam-guide-update-analysis]].
Strategic note
At 83% cold, founder could sit Cert 2 EARLY and bank the +$5k months ahead of the Nov-22 deadline β and it's the natural candidate for the "pass within first 3 months" escalator (by ~2026-08-26). Surfaced via iMessage 2026-06-14. βConfirm partner-portal gate is cleared (he has practice-exam access, so likely is β update the gate table when confirmed). 2026-06-27 caveat: exam 2 at 70% (borderline vs the 72% on record) tempers "sit it now" slightly β but the gap is ~3 cheap patterns (test-craft + hooks + API), not broad weakness, so a short targeted pass before sitting is the move. Re-confirm the real cut score either way.
Quiz log (appended by /cert-study quiz)
| Date | Cert | Scope | Result |
|---|---|---|---|
| 2026-06-10 21:32 ET | Anthropic CCA-F | D1 Γ3 (stop_reason, subagent isolation, hook enforcement) + D4 Γ1 (tool scoping/descriptions) + D3 Γ1 (Batches API) | 4/5 β D1 perfect (3/3), D4 perfect, D3 missed (Q5 Batches API, founder: "no clue"). Taught inline. Next quiz: D3-heavy (tool_choice modes, streaming events, extended-thinking budgeting). |
Related
- Skill:
~/.claude/skills/cert-study/SKILL.md - Milestones + negotiated comp lift:
~/rdco-vault/01-projects/phdata/milestones.md - Memory:
~/.claude/projects/-Users-ray/memory/project_phdata_cert_escalator_path.md - Snowflake study hub:
~/rdco-vault/01-projects/certifications/snowpro-genai-c02/ - Anthropic study hub:
~/rdco-vault/01-projects/phdata/anthropic-certification-study-guide.md
2026-08-24 β /cert-study cron: deadline correction + at-risk flag
Ran cron pulse. Today matches the stale pre-correction Snowflake deadline (2026-08-24) recorded in this file and in [[milestones]] line 17/24 β but per the founder's 2026-08-17 order-agnostic correction (already logged above under Cert 2's 2026-08-17 entries), that date is not the real Snowflake deadline. Corrected deadline is 2026-11-22. Updated the Cert 1 deadline line above; [[milestones]] still shows the stale 2026-08-24/90d framing and needs the same correction (not edited by this run β flagging for founder or next vault-health pass).
Verdict: Cert 1 (Snowflake) = at-risk. Zero study-log entries since job start (3 months), no exam scheduled, Track A sprint ([[study-plan]]) is only 7-10 days β so ~92 days of runway is still recoverable, but there is currently no plan in motion. Cert 2 (Anthropic) is done β passed 2026-08-17, escalator #1 paid.
This week's focus (Day 1 of Track A per [[study-plan]]): README.md + [[study-guide-summary]], log into Snowflake learning portal, download the official C02 study guide, register for the exam.
Founder forwarded a phData colleague's pass writeup (via iMessage, pasted). Full intel filed at [[2026-07-13-cca-colleague-pass-intel]]. Load-bearing items:
- Registration self-serve: Skilljar partner portal (phData email) β Pearson VUE (mandatory since Jul 1). The early-sit access gate finally has a concrete path.
- Difficulty map (60Q / 120min): HIGH = tool/error design for agents + escalation/HITL (scenario-heavy β matches founder's known D1 act-vs-surface weak axis from the 6/27 miss analysis). Biggest chunk = task decomposition (Med-High). Low-Med = session mgmt + agentic loop mechanics (founder's daily bread β skip drilling).
- Prep stack used: phData Udemy "Certified Claude Architect Masterclass 2026" + official docs + practice tests; elimination technique; third-party mock at claudecertificationguide.com/mock-exam (unvetted).
- Logistics: run Pearson VUE "test your system" well before exam day (colleague + others hit issues).
2026-08-31 β /cert-study cron pulse
Verdict: Cert 1 (Snowflake) = at-risk, second consecutive week. Still zero study-log entries, no exam scheduled, no track selected. 83 days remain to the corrected 2026-11-22 deadline β recoverable (>35d), but no plan is in motion and this is the same verdict as 2026-08-24. Cert 2 (Anthropic) remains done (passed 2026-08-17, escalator #1 paid, $200k base).
This week's focus (unchanged β Day 1 of Track A per [[study-plan]], still not started): README.md + [[study-guide-summary]], log into Snowflake learning portal, download the official C02 study guide, register for the exam.
iMessage sent per two-consecutive-weeks-at-risk trigger.
2026-09-07 β /cert-study cron pulse
Verdict: Cert 1 (Snowflake) = at-risk, third consecutive week. Still zero study-log entries, no exam scheduled, no track selected. 76 days remain to the 2026-11-22 deadline β still recoverable (>35d), but no plan is in motion and this is the same verdict as 2026-08-24 and 2026-08-31. Cert 2 (Anthropic) remains done (passed 2026-08-17, escalator #1 paid, $200k base).
This week's focus (unchanged β Day 1 of Track A per [[study-plan]], still not started): README.md + [[study-guide-summary]], log into Snowflake learning portal, download the official C02 study guide, register for the exam.
iMessage sent per at-risk-repeats-prior-week trigger (third consecutive week).
2026-09-14 β /cert-study cron pulse
Verdict: Cert 1 (Snowflake) = at-risk, fourth consecutive week. Still zero study-log entries, no exam scheduled, no track selected. 69 days remain to the 2026-11-22 deadline β still recoverable (>35d), but no plan is in motion; same verdict as the prior three weeks (08-24, 08-31, 09-07). Cert 2 (Anthropic) remains done (passed 2026-08-17, escalator #1 paid, $200k base).
This week's focus (unchanged β Day 1 of Track A per [[study-plan]], still not started): README.md + [[study-guide-summary]], log into Snowflake learning portal, download the official C02 study guide, register for the exam.
iMessage sent per at-risk-repeats-prior-week trigger (fourth consecutive week).
2026-09-16 15:10 ET β founder-initiated: starting Snowflake study, first real movement
Founder asked via iMessage (15:10) whether practice tests exist and what materials we have β the first concrete study action since the at-risk verdict began 2026-08-24 (now 4 consecutive weekly at-risk pulses, most recent 2026-09-14). Replied with the full materials inventory (README/study-guide-summary/study-plan/flashcards/5 domain files/supporting-materials.md) and the practice-test landscape: no free official Snowflake practice exam exists; third-party paid options (Udemy, Skillcertpro) are C01-labeled with unverified C02 currency; ExamTopics/brain-dump sites flagged as a cert-voiding policy violation per [[supporting-materials]]:153. Offered on-demand /cert-study quiz (scenario-based, graded, weak-area tracking) as the no-cost alternative. Asked whether he wants a quiz now or is starting at Day 1 of study-plan.md β awaiting his reply, not yet logged as an actual study session (no domain/topic engagement confirmed yet, just an information request).
67 days remain to the 2026-11-22 deadline as of this message.
2026-09-16 15:39 ET β high-signal colleague intel filed
Founder forwarded a phData colleague's (Kaulab Basu) internal Slack pass writeup for GES-C01 (passed weekend of 9/13-14) β a full 17-topic tiered study checklist with real exam-tested specifics. Filed verbatim + attributed to [[colleague-pass-checklist-kaulab-basu]] in the study hub, cross-linked from README.md and [[third-party-resources]] (now the #1 recommended read). Verified against our existing [[study-cortex-ai]] and [[study-governance]] before claiming novelty: genuinely new content is Cortex Analyst depth (VQR mechanics, multi-turn REST API fields, semantic-model relationship syntax, observability event tables), the exact Document AI privilege grant (CREATE SNOWFLAKE.ML.DOCUMENT_INTELLIGENCE at schema level), and the SPCS GPU_ compute-pool prefix requirement. Our RBAC/allowlist coverage (CORTEX_USER, CORTEX_MODELS_ALLOWLIST) already matched his account. This directly answered founder's "what about people sharing how the cert is, difficulty/challenges/topics" question from 15:12 β real answer, not the generic third-party links list.
2026-09-16 15:47 ET β 2 more colleagues' exam-experience notes filed
Founder forwarded 3 images (exam-overview slide, raw question list, rewritten/expanded version) β feedback from 2 more phData colleagues, question prompts only, no answers included. Filed to [[colleague-exam-experience-notes]], cross-linked in README. New surface vs. existing files: module_custom_instructions (Cortex Analyst topic-restriction), AI_PARSE_DOCUMENT's two failure modes (110MB size cap, page_split=TRUE + image conflict), external-app invocation of a Model Registry model. Deliberately framed as topic signal not an answer key β flagged to founder that literal predicted-question memorization would edge toward the brain-dump policy violation our own [[third-party-resources]] warns against, even though these specific images don't include answers.
2026-09-16 15:50-16:02 ET β recertification + account-consolidation research, both answered and sourced
Founder asked (15:50): (1) whether an existing SnowPro cert (email/level unspecified, expires December 2026) auto-renews via a higher-level pass, and (2) how to consolidate that cert record from ben@raydata.co onto bwilson@phdata.io. Dispatched research, verified against Snowflake's official CE program page (learn.snowflake.com/en/snowpro-continuing-education):
- CONFIRMED: "The highest certification renews all active ones below it... Specialty Certifications renews Core." SnowPro Specialty: Gen AI is explicitly listed as a covered program β passing GES-C02 before the Core cert's Dec 2026 expiration auto-renews it, no separate Core exam. Hard constraint: no date extensions, must land before expiration.
- Platform is CertMetrics (cp.certmetrics.com/snowflake), separate from the Learning/training account. No confirmed self-serve merge tool on Snowflake's tenant specifically (blocked from checking without login), but official guidance confirms mismatched learning/cert emails "cannot be automatically linked" and directs to a support case at community.snowflake.com/s/educational-services. Relayed both, with the fastest-path recommendation (check Profile for Request Merge first, file a case if absent). Also raised by founder in the same thread: colleague-sourced exam notes forwarded earlier may be C01-era ("version one") β already flagged in [[colleague-exam-experience-notes]]; confirmed to founder that C01βC02 is ~90% reusable per the folder's standing note, treating colleague content as topic signal not exact current syntax.
2026-09-21 β /cert-study cron pulse
Verdict: Cert 1 (Snowflake) = at-risk, fifth consecutive week (repeats 08-24, 08-31, 09-07, 09-14). Formal study-log table still has zero entries and no exam is registered β but real movement happened this week (2026-09-16 thread): materials inventory reviewed, two colleagues' pass writeups/exam-experience notes filed to the study hub, CE/recertification path confirmed, and a real registration-blocking risk (account-merge vs. partner-status) surfaced and is still open pending the founder's own CertMetrics login check. 62 days remain to the 2026-11-22 deadline β still recoverable (>35d). Cert 2 (Anthropic) remains done (passed 2026-08-17, escalator #1 paid, $200k base).
This week's focus: resolve the registration-account risk first (community.snowflake.com thread flagged 09-16, or file a CertMetrics merge case), THEN Day 1 of Track A per [[study-plan]] β README.md + [[study-guide-summary]], download the official C02 study guide, register for the exam. The colleague checklists ([[colleague-pass-checklist-kaulab-basu]], [[colleague-exam-experience-notes]]) are now the highest-signal starting material β lead with those over the generic Day 1 sequence.
iMessage sent per at-risk-repeats-prior-week trigger (fifth consecutive week).
2026-09-16 16:14-16:26 ET β partner-status vs. account-merge risk, flagged before registration
Founder updated CertMetrics "partner status" from Mammoth Growth to phData under his ben@raydata.co profile (no merge-account option found there) and asked whether that's sufficient before registering the new exam under his RDCO email. Flagged a real risk immediately (before research completed): if the new exam is registered under a different email/profile than the one holding his existing Core cert, the accounts may not link and the auto-renewal ("highest cert renews all below it") could fail to fire, risking the Dec 2026 Core expiration. Research confirmed no official Snowflake doc defines "partner status" directly; cross-vendor CertMetrics pattern (AWS/Atlassian/Google Cloud) suggests it's attribution metadata separate from account identity (inferred, not Snowflake-verified). Renewal-linkage risk itself also lacks direct documentation but is a reasoned inference from Snowflake's own "accounts cannot be automatically linked" language + existence of a dedicated merge-case process. Recommendation relayed: register/sit the Gen AI exam under the SAME ben@raydata.co login as Core; keep partner-status update as-is (separate, lower-stakes). Surfaced one directly-on-point but JS-rendered community thread he can check with his own login for an authoritative answer: community.snowflake.com/s/question/0D5Do00001360WyKAI/adding-certification-to-the-partner-portal
- 2026-09-23 11:05 ET: founder back to studying (iMessage: "I'm going to get back to studying"). Study session in progress.
- 2026-09-23 13:46 ET: EXAM BOOKED: GES-C02 Monday 2026-09-28 at 2 PM (founder via iMessage; timezone presumably ET, not stated). He'll cram Mon morning. He re-sent the official study guide PDF; it's byte-identical (sha1 0ef1aea7) to source-materials/SnowProGenAIStudyGuideC02.pdf from 05-18. | 2026-09-24 12:0x ET | SnowPro GenAI GES-C02 | quiz resumed at Q4: Cortex Search Γ3 (TARGET_LAG, ON clause vs ATTRIBUTES, SEARCH_PREVIEW) βββ; running 6/7 incl. 09-23 Q1-Q3b (miss: AI_COMPLETE required args). Exam Mon 9/28 2pm stands (founder 09-24 08:49). | Next: Cortex Analyst | | 2026-09-24 14:5x ET | SnowPro GenAI GES-C02 | Q10 governance β (docs now: USE AI FUNCTIONS + CORTEX_USER/AI_FUNCTIONS_USER) Β· Q11 AI_PARSE_DOCUMENT LAYOUT β Β· Q12 FINETUNE prompt/completion β Β· Q12b FINETUNE('DESCRIBE') β β fine-tuning = weak area, re-drill Mon AM | running 11/14 | | 2026-09-24 18:3x ET | SnowPro GenAI GES-C02 | Q13 AI_FILTER β β Q13b AI_CLASSIFY β Β· Q14 Agents (Analyst+Search) β Β· Q15 guardrails β (picked response_format) β 2nd weak area: AI_COMPLETE optional args (with 09-23 Q3); Mon cram = fine-tuning + AI_COMPLETE arg map | running 13/18 | | 2026-09-24 19:5x ET | SnowPro GenAI GES-C02 | Q15b model_parameters β Β· Q16 AI_EXTRACT vs PARSE β (picked PARSE OCR; his VARIANT objection was valid, question wording was sloppy) | running 14/20. Mon cram: fine-tuning, AI_COMPLETE arg map, EXTRACT vs PARSE vs FILTER vs CLASSIFY. Strong: Cortex Search, Analyst, Agents, governance |
- 2026-09-28 11:34 ET: exam-day cram, founder: "Let's get back into quiz mode." Resumed at Q17 (running 14/20). Order: fine-tuning β AI_COMPLETE arg map β EXTRACT/PARSE/FILTER/CLASSIFY. Current docs (FINETUNE('CREATE') page, checked 11:4x): base_model currently only 'llama3.1-8b' (older exam material may list mistral-7b/llama3-8b); training + validation queries need
prompt+completioncolumns; validation query + options (max_epochs 1-10) optional; returns job ID string; model saved to the schema's model registry. | 2026-09-28 11:3x-12:3x ET | SnowPro GenAI GES-C02 | Q17 FINETUNE prompt/completion β Β· Q18 FINETUNE('DESCRIBE') β (was a 09-24 miss) Β· Q19 tuned-model inference = AI_COMPLETE + USAGE on model β β fine-tuning CLEARED Β· Q20 AI_COMPLETE arg map β (put 'usage' in model_parameters; token counts = show_details TRUE) β Q20b structured_output location | running 17/24 | | 2026-09-28 12:3x-13:3x ET | SnowPro GenAI GES-C02 | Q20b structured_output β Β· Q21 EXTRACT/FILTER/CLASSIFY/PARSE(LAYOUT) match β Β· Q22 AI_CLASSIFY output_mode multi β Β· Q23 AI_EXTRACT file+table schema β Β· Q24 CORTEX_ENABLED_CROSS_REGION β Β· Q25 CORTEX_MODELS_ALLOWLIST β β all three weak areas cleared except AI_COMPLETE arg map (1 miss Q20, then β) | running 23/30 (77%); exam-day session 9/10 after Q20 miss. Docs notes: cross-region default now ANY_REGION for new commercial accts (exam key likely DISABLED); ALLOWLIST default 'All'. Sent review sheet 13:3x. | - 2026-09-28 13:32 ET: founder: "15 minutes before I start trying to log in. You think I'm ready?" + completed the Udemy course, found the quiz more useful. Ray: yes (77%/30, 9 of last 10, weak areas turned; 750 scaled cut = C01 proxy, C02 unconfirmed). Last-minute skim sent: Cortex Search CHANGE_TRACKING/serverless, Analyst VQR 3-vs-5 + multi-turn messages, REVOKE CORTEX_USER FROM PUBLIC + two-layer access. Exam 2pm; result pending.
- 2026-09-28 15:39 ET: β PASSED SnowPro Specialty: Gen AI (GES-C02) β founder iMessage with Snowflake credential email screenshot ("Benjamin Wilson, here is your credential for SnowPro Specialty: Gen AI Certification"). Both phData cert escalators now banked (Anthropic CCA-F 2026-08-17, Snowflake GenAI 2026-09-28). Founder: "The test was much harder and different than what I studied for." Also completed a Udemy course; said Ray's quiz helped more. Ray asked for: what was different (feeds a C02 checklist for phData colleagues + first LinkedIn post) and confirm CertMetrics shows it on the same profile as SnowPro Core (auto-renew link). Score not shared.
- 2026-09-28 15:44 ET: founder confirmed SnowPro Core expiry now 2028-09-28, matching the Specialty (auto-renew worked, same CertMetrics profile). Exam shape: multi-select A-F "best 2-3", no syntax drills (some parameters), out-of-the-blue SPCS custom Docker image question. Filed: [[founder-exam-experience-2026-09-28]]. Offered LinkedIn post #1 draft ("studied for the wrong exam and passed anyway"), awaiting his yes.
- 2026-09-28 15:49 ET: SCORE: 854 (founder found it post-exam). His original SnowPro Core was 822. Ray: strong pass, ~100 over the 750 scaled cut (750 = working proxy, C02-specific cut not officially confirmed); no public percentile data. Parallel: Anthropic CCA-F 850/1000 (cut 720) on 2026-08-17.
2026-10-12 β /cert-study cron pulse
Verdict: both certs DONE. Cert 2 (Anthropic CCA-F) passed 2026-08-17 (850/1000), escalator #1 paid same day ($200k base). Cert 1 (Snowflake GES-C02) passed 2026-09-28 (854), auto-renewed SnowPro Core. No study-tracking action remains for either cert.
Open item (not a study gap β a paperwork gap): Unlike escalator #1, this log has no entry confirming the Snowflake cert was submitted to phData for escalator #2 (+$5k, base β $205k). 14 days have passed since the 09-28 pass with no logged submission. Flagging via iMessage rather than assuming either way.
2026-10-19 β /cert-study cron pulse
Verdict: both certs DONE, unchanged from 2026-10-12. Cert 2 (Anthropic CCA-F) passed 2026-08-17 (850/1000), escalator #1 paid same day. Cert 1 (Snowflake GES-C02) passed 2026-09-28 (854), auto-renewed SnowPro Core. No study-tracking action remains for either cert; no formal cron trigger (at-risk/off-track/stale/blocked/unscheduled-exam) applies.
Open item unchanged: escalator #2 (Snowflake, +$5k β $205k base) paperwork-submission confirmation is still not logged β 21 days since the 09-28 pass, 7 days since the 10-12 iMessage nudge, no reply logged. Not re-pinging this week (nudged 7 days ago; repeat-weekly pings on the same unconfirmed item risk noise per standing signal-to-noise guidance) β will re-flag if still open at the next pulse (2026-10-26) or if the gap widens further.
2026-10-21 β AWS certs (phData push)
Founder asked (07:14 ET) which AWS certs to pursue and in what order. Cert targets also feed the 2026 review-cycle goals ([[2026-10-21-2026-review-goals-draft]], [[2026-goals-ben-wilson]]). Ray rec (sent iMessage): finish CCDV-F first, then (1) AWS AI Practitioner (quick, overlaps GenAI/CCA-F prep), (2) Solutions Architect β Associate (recognized, AWS plumbing), (3) Generative AI Developer β Professional (flagship fit: agents/RAG on Bedrock). Later optional: Data Engineer β Associate or ML Engineer β Associate. Skip Cloud Practitioner (covered) and ML Specialty (retiring). Ask phData: which certs count toward partner tier; voucher availability. Lineup from vendor/training-site search (trainingcamp.com, cloudthat.com), NOT yet verified on aws.amazon.com: verify codes/status before he books. Each extra cert = $500 bonus + deal placement (memory project_cert_bonus_and_deal_placement).
07:18 founder: will do "both Claude ones" first; agrees with the AWS order (AI Practitioner β SA Associate β GenAI Developer Pro). All count toward phData partner tier; discounted vouchers; phData reimburses first two attempts + pass bonus. Ray verified on docs.aws.amazon.com: GenAI Developer β Professional = AIP-C01, live, 65 scored Qs, 5 domains (FM integration/data/compliance 31%, implementation 26%, safety/governance 20%, ops efficiency 12%, testing 11%), model training out of scope, suggested 2+ yrs AWS. Asked which second Claude cert he means (CCA-F already passed 08-17); awaiting.
07:19 founder: has Claude Architect Foundations; next Claude Developer Foundations, then Claude Architect Professional. Asked for other Snowflake certs. Ray rec (from learn.snowflake.com listing, exam guides not read): SnowPro Advanced: Architect (ARA-C01) top pick, slot after Claude Architect Pro and before the AWS three; MLOps Engineer (Advanced) = maybe, scope unverified, Ray owes an exam-guide check; skip Snowpark Specialty + Advanced Data Scientist (site says both retire 2026-11-01) and DE/Analyst/Admin/Security (off-lane). Suggested asking his lead which partner tier is nearest its threshold. Proposed full sequence: CCDV-F β Claude Architect Pro β SnowPro Adv Architect β AWS AI Practitioner β AWS SA Associate β AWS GenAI Dev Pro.
07:23 founder PLAN (his words): Claude certs through end of 2026; AWS Q1 2027 ("or maybe one by the end of the year"); Snowflake early Q2 2027, because "I just passed the cortex cert, so passing another Snowflake one right now wouldn't extend those certs for much longer" (founder correction of Ray's slot-Snowflake-before-AWS suggestion). Ray-drafted target dates (awaiting his OK): Claude Developer Foundations mid-Nov Β· Claude Architect Professional (CCAR-P; live, 120 min, 720 pass, $175, partner-network only per third-party sources; scope claims unverified) mid-Dec Β· stretch AWS AI Practitioner last week of Dec Β· AWS SA Associate Jan-Feb Β· AWS GenAI Dev Pro March Β· SnowPro Advanced Architect Apr-May 2027 (MLOps Engineer only if exam guide checks out). Offered: calendar target dates + retune weekly cert pulse to this plan.
07:24 founder 'Yes please' β 6 all-day FREE target events created on ben@raydata.co (11-20 CCDV-F, 12-16 CCAR-P, 12-30 AWS AI Practitioner stretch, 2027-02-26 AWS SAA, 2027-03-26 AWS AIP-C01, 2027-04-30 SnowPro Adv Architect). /cert-study SKILL.md rewritten to the plan (backup SKILL.md.bak-2026-10-21-plan): escalator deadlines + partner-portal gate removed, plan table + vendor-source rule + no-exam-content rule added, gotchas kept (date first, length-matched distractors, no babysitting, append-only log).
2026-10-21 14:30 ET: Founder screenshot (Anthropic cert portal banner): Associate-Foundations, Developer-Foundations and Architect-Professional exams get updated content starting 2026-12-07; updated exam guides available 2026-11-17. Ray rec: take CCDV-F on the CURRENT content (keep 11-20 target, must sit before 12-07; leaves ~2.5 weeks of retake buffer); CCAR-P stays on the NEW content (study from the 11-17 guide, keep 12-16 target, slide to late Dec is fine). Don't cram CCAR-P before 12-07. Awaiting founder's call.
2026-10-21 14:41 ET: Founder re-sequence (iMessage 14:33-14:40): CCDV-F 11-10 (current content) β AWS AI Practitioner 12-04 β CCAR-P 12-22 (new content, prep from 11-17 guide). Q1 AWS dates unchanged. May pull forward if CCDV-F clears early. Calendar events moved (ben@raydata.co). Goals doc KR 4.1 stays 3 Claude certs in 2026; KR 4.3 AWS AI Practitioner added. The reasoning behind sitting CCDV-F before 12-07 and holding CCAR-P for the new guide is re-examined, and partly corrected, in [[2026-11-07-ccdvf-recertification-exposure-2026-12-07-content-change]].
2026-10-20 β CCDV-F readiness + study pack built
Same-day cold-sit assessment, study resources, fact sheet, and a 50-question gap-weighted quiz, built from the exam guide PDF the founder pasted: [[2026-10-20-ccdv-f-readiness]] (domain-by-domain readiness, ~55% cold / ~85%+ after ~10h prep), [[2026-10-20-ccdv-f-study-resources]] (ranked resource list against the readiness note's prep plan), [[2026-10-20-ccdv-f-study-sheet]] (red/yellow fact sheet, doc-checked), [[2026-10-20-ccdv-f-quiz-1]] (the quiz itself), and [[2026-10-20-ccdv-f-mso-course-notes]] (founder-pasted MSO Foundations course screens with calibration notes). Weak-area pattern matches the CCA-F history above: raw API surface (Message Batches, tool_choice, streaming, caching) stays the gap daily Claude Code use never drills.
2026-10-27: CCDV-F iMessage quiz rounds 1-2 (one question at a time)
Round 1 (10-26 18:44 β 10-27 06:10 ET, key [[2026-10-26-ccdv-f-imessage-quiz-2-key]]): 6/8. Misses: pause_turn (server vs client tools), MCP primitives (tool/resource/prompt by who controls it).
Round 2 (10-27 06:11-06:21 ET, mixed topics, founder: "not weight too much... getting more reps in"): 7/8. Miss: cache prefix order (tools β system β messages; a tool change invalidates the whole prefix). Correct: repo CLAUDE.md, tool_result in user turn, subagent context isolation, tool descriptions, Streamable HTTP, evals, .claude/settings.json allow rule.
Combined 13/16. Founder asked on 10-26 whether to wait for the December exam refresh; Ray rec = sit 11-10, his call, no decision yet.
2026-10-27 (founder iMessage ~14:57 ET): CCDV-F official practice exam (Practice mode Beta), attempt 1 of 6 available: 84% (45/53), 1h40m, light study. Domains: Applications & Integration 76% (17q, ~4 missed), Model Selection & Optimization 78% (9q, 2 missed), Security & Safety 75% (4q, 1 missed), Agents & Workflows 88% (8q, 1 missed), Tools & MCPs 100% (6), Prompt & Context Eng 100% (6), Claude Code 100% (2), Eval/Testing/Debugging 100% (1). Founder sending missed questions for review; next quiz reps weight Apps & Integration + Model Selection.
- Misses debriefed (founder: "between two and chose the wrong side"): Q7 Models API vs probing usage; Q9 client executes tools (picked "API runs client tools"); Q18 effort = soft + broad (picked "low guarantees no thinking"); Q26 structured outputs vs prefill brace; Q27 resend full history (stateless) vs max_tokens caps history; Q31 top-level system vs role:system first message; Q35 content_block_delta vs message_delta; Q37 canonical path resolution vs "../" substring filter. Patterns: (1) hack vs dedicated feature/code control, (2) stateless + client-side execution, (3) soft vs hard (only max_tokens is hard), (4) exact event names. Offered a 6-q targeted round.
- 15:06 founder asked readiness + tomorrow vs Friday. Ray rec: Friday 10-30; ready = 2 more fresh practice exams Wed/Thu at 85%+ each, Apps & Integration 80%+; Thu OK if Wed 88%+; <80% -> slip a week (still ahead of 11-10). Pass mark not in notes: he checks portal. Awaiting his call.
- 15:10 founder: passing mark 720/1000 scaled on all the exams ("pretty sure"); question count/weighting differs per exam. Plan unchanged (Friday if next 2 practice >=85%).
2026-10-30: CCDV-F (Claude Certified Developer - Foundations) PASSED 926/1000 (bar 720). 100% on 21/25 sections; 0% Agent Construction, Understanding Reqs, MCP Dev; 67% Claude App Design. Report for $500 bonus. Next: AWS AI Practitioner 12-04. Nominal expiry ~2027-10-30 (12-month validity); whether the 2026-12-07 content change puts this credential under a full-retake-to-recertify risk is researched in [[2026-11-07-ccdvf-recertification-exposure-2026-12-07-content-change]] (answer: permissive clauses exist in three Anthropic documents, none yet exercised β watch the 2026-11-17 guide's version-vs-revision labeling).
2026-10-30 13:50 ET: Founder posted the CCDV-F pass on LinkedIn (his line: "...which is exactly what I get to help our clients do."). He asked whether to do CCAR-P now or AWS first. Ray rec: AWS AI Practitioner pulled forward to ~11-20, then CCAR-P 12-22 on the new content (guide 11-17, content change 12-07). Asked whether to move the calendar target, and whether the LI post counts as Ship 30 post 2. Awaiting.
2026-10-30 16:17 ET: Founder: phData AWS lead endorsed going straight to AIP-C01 (skip AI Practitioner, roll back if in too deep). Shared a Glean study pack (official guide and Skill Builder; phData WorkRamp course, AI Skills Club deck, Ryan Leonard advice, beta-exam pointers: Guardrails, cross-Region, RAG chunking, AppConfig, Step Functions, latency, observability, evals; third-party content reported dated). Ray agreed, recommended a cold official pretest in week 1 as the rollback gate, and is building the study folder aws-aip-c01/. Calendar: the 12-04 event is now an AIP-C01 placeholder (TBD after the pretest); the 2027-03-26 slot is open. cert-study SKILL table updated. Internal links are pending from the founder.
2026-10-30 16:35 ET: aws-aip-c01/ study folder built (sw-builder) and gated (sw-critic: REVISE on phone usability only; fix applied by Ray with a count==1 check). Sent to the founder: pretest-first, A/B/C miss tags, rollback rule (<60% and A_share>=40% -> AI Practitioner; thresholds are Ray-chosen). phData internal links pending.
2026-11-02 β /cert-study cron pulse
Verdict: on-track. Current cert = AWS AIP-C01, target 2026-12-04 placeholder (32d out). [[study-plan]] Phase 1 ("week 1: official pretest, cold") window runs through ~11-06; the Pretest log table (study-plan.md line 56) is still blank β pretest not yet taken, which is expected at day 3 of that window, not a miss. A 60-question [[quiz-bank]] (Domains 1-5 + F1-F15 AWS-fundamentals appendix) was built 2026-11-01, ahead of the pretest β per the plan's own instruction ("take the pretest before reading anything in this folder"), it should stay unopened until after the cold pretest is scored, so it doesn't contaminate the rollback read.
This week's focus: take the AWS Certification Official Pretest cold via the Skill Builder exam-prep page (resources.md line 20, skillbuilder.aws/category/exam-prep/generative-ai-developer-professional-AIP-C01 β step 2 "Official Pretest"), no lookups during it, then score misses into the A/AWS-foundations Β· B/Bedrock-mechanics Β· C/GenAI-design buckets from study-plan.md lines 32-38 and fill the Pretest log table (line 56-66) β that score + A_share is what the pre-registered rollback rule (lines 42-54) runs on. Next cert (CCAR-P) starts prep from the new guide ~2026-11-17, target 2026-12-22 (50d out).
iMessage: not sent (on-track, 32d to target, exam not yet a booking question until the pretest resolves stay-vs-rollback).
2026-11-02/07 β CCAR-P content-delta and CCDV-F recertification exposure researched
[[2026-11-02-claude-architect-pro-2026-12-07-content-delta]] verifies the pre-change CCAR-P v1.0 baseline (7 domains, 63 items, 720/1000) and finds Agent SDK coverage is currently zero while Skills and MCP are one objective apiece β confirming ~61% of the exam (Solution Design, Evaluation/Testing, Governance/Safety, Stakeholder Comms) is version-independent and safely studiable now. It recommends keeping the 12-22 sit date over squeezing in before 12-07, archives the guide PDF locally (S3 link is unversioned), and flags a hard 2026-11-17 re-diff as a scheduled task, not a hope.
[[2026-11-07-ccdvf-recertification-exposure-2026-12-07-content-change]] follows up on whether the same 12-07 change threatens the already-passed CCDV-F credential (line 238 above): yes on paper (identical Β§14 language plus a Terms-and-Conditions clause the parent brief missed), no in practice (all three clauses are permissive, none ever exercised, and the public FAQ concedes "full details will be shared before the first certifications come up for renewal"). No action needed; re-check the 2026-11-17 guide's Document Control table for "new version" vs. "revision" language.
2026-11-09 β /cert-study cron pulse
Verdict: off-track. Current cert = AWS AIP-C01, target 2026-12-04 placeholder (25d out). [[study-plan]] Phase 1's cold-pretest window ran through ~11-06; it is now 11-09 with the Pretest log table (study-plan.md lines 56-66) still blank β no pretest taken, no study-log entry since the 11-02 pulse. This has also been sitting unaddressed on the morning-prep homework rail since at least 11-03. The rollback rule (AIP-C01 vs fallback to AI Practitioner) and the real exam date both depend on this score, so until it's taken the Dec-4 date stays a placeholder, not a booking.
This week's focus: take the AWS Certification Official Pretest cold via Skill Builder (resources.md line 20, skillbuilder.aws exam-prep page β step 2 "Official Pretest"), no lookups during it, then bucket misses A/AWS-foundations Β· B/Bedrock-mechanics Β· C/GenAI-design per study-plan.md lines 32-38 and fill the Pretest log table (lines 56-66) β quiz-bank.md stays unopened until that's done, per its own header note. Next cert (CCAR-P) starts prep from the new guide ~2026-11-17, target 2026-12-22 (43d out).
iMessage: sent β off-track trigger (behind the plan's own pace, β€28d to target).
2026-11-16 β /cert-study cron pulse
Verdict: off-track (2nd week running). Current cert = AWS AIP-C01, target 2026-12-04 placeholder (18d out, exam not booked). Pretest log (aws-aip-c01/study-plan.md) still blank; no founder progress entry since 11-09 (no pretest signal on the iMessage thread either). Rollback rule and the real exam date both still hinge on the cold pretest. Founder's Foodsmart week is heavy; if the pretest can't happen this week, a re-date of 12-04 is the honest move (ask once).
This week's focus: take the Official Pretest cold (Skill Builder exam-prep page, resources.md line 20), no lookups, bucket misses A/B/C into study-plan.md Pretest log; quiz-bank.md stays closed until then. Next: CCAR-P prep from the new guide (published 2026-11-17, re-diff scheduled), target 2026-12-22 (36d out).
iMessage: attempted, FAILED (AppleEvent -1712 timeout); saved to cron-undelivered for the main session.