Local Agentic AI Masterclass with Gemma 4 & Apple MLX
Build StockAgent: A 100% local, zero-API-cost autonomous financial research copilot using Gemma 4, Apple MLX, and Pydantic
Course Syllabus
4 progressive hands-on modules designed for local implementation with zero cloud API costs.
Lab 1: Run Gemma 4 (12B) Locally with Apple MLX & Token Streaming
Build the local inference foundation on Apple Silicon Metal GPU with zero-copy unified memory loading, real-time token streaming, and zero API costs.
Lab 2: Type-Safe Pydantic Tool Calling & Live Temporal Grounding
Eliminate date hallucination and stale training cutoffs by binding Gemma 4 to strict Pydantic v2 schemas and live Google Serper market search.
Lab 3: Quantitative Valuation Multiples & Penny Stock Purge Engine
Implement live market table scraping, enforce the Penny Stock Purge ($5 floor / $1B cap), calculate valuation multiples, and generate executive investment memos.
Lab 4: StockAgent — Autonomous Financial Market Copilot
Assemble the complete autonomous market screener copilot, integrating MLX local inference, Pydantic tool calling, and live Yahoo Finance scraping.