SnowPro Gen AI C02 — interactive study guide

Domain-by-domain walkthrough of the official Snowflake C02 study guide. Click any domain to expand its objectives, subtopics, and supporting-material cross-links. Resource URLs go to Snowflake docs or community blogs.

→ practice quiz study plan supporting materials index

domain weighting

1.0 18% 2.0 38% 3.0 29% 4.0 15%

Community wisdom: Domains 2 and 3 are heavily over-indexed on the exam vs. their weighting suggests. Spend extra time on Cortex function discrimination (D2) and RBAC + cost (D3).

Domain 1.0 Snowflake for Gen AI Overview 18%

Foundational map of what Snowflake offers for Gen AI: Cortex AI surface, interfaces, BYOM via Model Registry + SPCS, and Snowflake Intelligence as the higher-level chat layer.

objectives

1.1 Define Snowflake's Gen AI principles and features
  • Snowflake Cortex
  • Cortex Models and Functions
  • Cortex Fine-tuning (Public Preview)
  • Cortex Search — RAG + unstructured-data use cases
  • Cortex Analyst — text-to-SQL use cases
  • Cortex Agents
  • Snowflake Cortex Code (Snowsight UI + CLI)
  • Snowflake Copilot Inline (Public Preview)
  • Snowflake Intelligence
  • Different interfaces — AI Studio, SQL, REST API
  • Bringing your own models into Snowflake
  • Snowflake Model Registry (custom model)
  • Snowpark Container Services
1.2 Outline Gen AI capabilities in Snowflake
  • Prompting
  • Cortex AI functions — Vector-embedding, Context Windows
  • Cortex Search — Multi-index queries, access control requirements, different ways to use
  • Cortex Analyst — Semantic Views, Semantic Views Autopilot, YAML Specification, Verified Query, Custom Instructions
  • Cortex Agents
  • Snowflake Intelligence
  • Cross-region inference — CORTEX_ENABLED_CROSS_REGION parameter; latency + availability considerations
  • REST APIs
  • Model Context Protocol (MCP)
  • Snowflake Cortex Code — CLI commands
  • Cortex Knowledge Extensions (CKE)

study resources (Snowflake docs + blogs)

20 resources — show

vault study notes

Domain 2.0 Snowflake Gen AI Functions 38%

LARGEST domain (38%). Function-by-function mastery: AI_* + legacy SNOWFLAKE.CORTEX.* + helpers + vector functions. Hands-on patterns for data extraction, enrichment, augmentation, RAG, text-to-SQL, and third-party-model deployment via SPCS + Model Registry.

objectives

2.1 Apply AI functions in Snowflake
  • Snowflake Cortex AI functions — general
  • AI_COMPLETE, COMPLETE (with Structured Outputs)
  • Task-specific: AI_CLASSIFY, AI_EXTRACT, AI_PARSE_DOCUMENT, AI_SENTIMENT, SUMMARIZE, AI_SUMMARIZE_AGG, AI_TRANSLATE, AI_EMBED, AI_FILTER, AI_AGG, AI_SIMILARITY, AI_TRANSCRIBE, AI_REDACT
  • Vector functions: VECTOR_INNER_PRODUCT, VECTOR_L1_DISTANCE, VECTOR_L2_DISTANCE, VECTOR_COSINE_SIMILARITY, VECTOR_TRUNCATE, VECTOR_NORMALIZE, VECTOR_SUM/MIN/MAX/AVG
  • Helpers: AI_COUNT_TOKENS, TRY_COMPLETE, SPLIT_TEXT_RECURSIVE_CHARACTER, SPLIT_TEXT_MARKDOWN_HEADER, TO_FILE, PROMPT
2.2 Perform data analysis given a use case
  • Fully-managed LLMs, RAG, and text-to-SQL services
  • Unstructured data — AI_PARSE_DOCUMENT, AI_EXTRACT, AI_SIMILARITY, AI_COMPLETE
  • Cortex Search — recursive split text markdown, chunk sizing, embedding models, semantic reranking
  • Multi-modal — Audio + Image processing
  • Structured data — AI_COMPLETE + Cortex Analyst (Verified Query Repository, integration with Cortex Search, Suggested Questions, CUSTOM_INSTRUCTIONS)
  • Performance considerations — choosing a model, latency vs accuracy, fine-tuning, model capability, Provisioned Throughput
2.3 Build or interact with interfaces to chat with data
  • Set up the Snowflake environment — required privileges
  • Invoke Cortex functions in application code (e.g. Streamlit in Snowflake)
  • Chat conversations — multi-turn architecture, update parameters (messages array for conversation history)
  • Snowflake Intelligence
2.4 Apply Snowflake Cortex functions in data pipelines
  • Snowflake Cortex via SQL interface
  • Data extraction, enrichment, augmentation, transformations
2.5 Run third-party models in Snowflake
  • Snowpark Container Services — environment setup, Docker images, specification files, compute pools, image repositories
  • Snowflake Model Registry — logging the model, calling the model

study resources (Snowflake docs + blogs)

32 resources — show

vault study notes

Domain 3.0 Snowflake Gen AI Governance 29%

Community-flagged "make-or-break" domain. RBAC + privileges, model-access controls, cost governance across Cortex AI surfaces, AI observability via TruLens, and the ACCOUNT_USAGE views that show what your org is actually spending.

objectives

3.1 Set up model access controls
  • Limits on which models can be used
  • Restrict access to specific models — application roles
  • Control model access — RBAC, account-level allowlist parameter
  • Data safety + security: Cross-region inference, Guardrails, Sensitive data management (AI_REDACT)
  • Methods to reduce model hallucinations and bias
  • REST API authentication methods
3.2 Grant and revoke RBAC and privileges
  • Individual privileges — specific requirements for Analyst, Search, Agents, Snowflake Intelligence
  • Roles: CORTEX_USER, CORTEX_ANALYST_USER, CORTEX_AGENT_USER, CORTEX_EMBED_USER
3.3 Manage, monitor, and optimize Snowflake Cortex costs
  • Cortex Agents — limit token usage
  • Cortex Search — virtual warehouse, EMBED_TEXT, serving, indexing (3 cost components)
  • Cortex Analyst
  • Cortex AI functions — minimize tokens, token cost implications
  • Tracking SPCS costs — compute pools
  • Tracking model usage + consumption — usage quotas + ACCOUNT_USAGE views: CORTEX_ANALYST_USAGE_HISTORY, CORTEX_AISQL_USAGE_HISTORY, CORTEX_SEARCH_DAILY_USAGE_HISTORY, CORTEX_REST_API_USAGE_HISTORY, CORTEX_PROVISIONED_THROUGHPUT_USAGE_HISTORY, METERING_DAILY_HISTORY, METERING_HISTORY
  • Object tagging to monitor AI services costs
3.4 Use Snowflake AI observability tools
  • Evaluation metrics
  • Comparisons
  • Tracing
  • Logging + Event tables
  • Implementation methods — TruLens SDK

study resources (Snowflake docs + blogs)

34 resources — show

vault study notes

Domain 4.0 Snowflake Document Processing 15%

Smallest by weight, but every question is concrete: AI_PARSE_DOCUMENT modes, AI_EXTRACT response formats, pipeline orchestration with Streams + Tasks, and troubleshooting GET_PRESIGNED_URL access errors.

objectives

4.1 Use document parsing functions
  • AI_PARSE_DOCUMENT — OCR mode, LAYOUT mode, page_split, page_limit
  • AI_EXTRACT — response format, prompt engineering
4.2 Prepare and manage documents and implement extracting workflows
  • Upload documents
  • Requirements — formats, size limits
4.3 Build automated document processing pipelines with Cortex AI integration
  • Orchestration — Streams, Tasks
4.4 Troubleshoot and optimize document processing
  • Extracting query errors — GET_PRESIGNED_URL function
  • Requirements and privileges
  • Cost and best-practice considerations
  • Fine-tuning arctic-extract models

study resources (Snowflake docs + blogs)

11 resources — show

vault study notes

how to use this guide

  1. Read the official PDF first — at ~/rdco-vault/01-projects/certifications/snowpro-genai-c02/source-materials/SnowProGenAIStudyGuideC02.pdf
  2. For each domain, expand the objectives and verify you can describe each subtopic in 1-2 sentences without looking. Gaps → click into the linked Snowflake doc.
  3. Pair this guide with study-cortex-ai.md for the function-table density needed on the exam.
  4. When you can answer 80%+ on the practice quiz across two attempts in a row, you are ready.
  5. Post-exam: write post-mortem-c02-attempt-N.md in the cert folder.