SnowPro Specialty: Gen AI Certification
Design, build, and deploy production-grade Generative AI, LLM applications, retrieval-augmented generation (RAG), and autonomous conversational agents using Snowflake Cortex AI.
Official Exam Specification & Quick Facts
Question Difficulty Breakdown
Exam Guidelines & Simulation Rules
Frequently Asked Questions (FAQ)
The exam features 65 scenario, SQL, and LLM architecture questions over 115 minutes. Cost is $375 USD with a passing threshold of 750/1000 (75%). SnowPro Core is a required prerequisite.
Domains include: (1) Snowflake Cortex AI Architecture (~25%), (2) LLM Functions and Prompt Engineering (~25%), (3) Vector Search and Retrieval-Augmented Generation (~25%), and (4) Governance, Security, and Fine-Tuning (~25%).
Snowflake Cortex AI is a fully managed, serverless AI service that provides instant access to leading foundation models (Llama, Mistral, Gemma) and machine learning functions directly inside SQL with zero infrastructure.
Core functions include: `COMPLETE()` (text generation), `EXTRACT_ANSWER()` (reading comprehension), `SENTIMENT()` (sentiment scoring), `SUMMARIZE()` (text summarization), and `TRANSLATE()` (language translation).
Cortex Search provides a fully managed, hybrid search service (vector embeddings + keyword BM25) over unstructured enterprise documents, delivering sub-second retrieved context to foundation models.
The native `VECTOR` data type stores dense mathematical embeddings (e.g. `VECTOR(FLOAT, 768)`). It integrates with vector distance functions (`VECTOR_COSINE_SIMILARITY`) for similarity search.
Cortex Fine-Tuning allows organizations to customize open foundation models (e.g. Meta Llama) using domain-specific prompt-response training data directly within their Snowflake security perimeter.
Cortex Analyst provides conversational text-to-SQL capabilities over structured enterprise schemas. Cortex Search provides hybrid lexical and semantic search over unstructured documents.
Customer data used with Cortex AI never leaves the Snowflake security perimeter, is never used to train third-party foundation models, and is protected by Snowflake's RBAC and encryption governance.
Streamlit in Snowflake allows developers to build and share interactive, Python-based web applications that call Cortex LLM functions and display visualizations directly inside Snowflake without hosting external servers.
The credential is valid for two years from the date earned.
This specialty certifies Generative AI Engineers, Enterprise AI Architects, and LLM Application Specialists earning average US salaries between $155,000 and $215,000+.