Sample projects I’ve built in Python, and with visual agent builders like Google Opal.
📋 Table of Contents
A churn-prediction agent built in Google Opal and scored against evals. Developed for a hands-on agent build series I co-hosted for an enterprise SaaS client and MIT Sloan MBAs.
Three templated builds increase in complexity, from relying on Opal agents to choose how to analyze a dataset, to running specific code the user generates, to enabling Opal to write and execute its own code.
Agent that generates and executes its own code to predict churn risk
Follow-along build guide:
Google Opal – RevOps Churn Agent
A tool-using weather agent built with the Anthropic SDK to learn the agent loop from the ground up: tool use, error handling, iteration.
https://github.com/laoudji/weather-agent
SDKs aimed at observing agents, evaluating them, and enabling trust.
https://github.com/stars/laoudji/lists/ai-safety-observability
Toolkits for everything from controlling liquid handlers (shoutout PyLabRobot, featured guest on Discovery Engines), to fine-tuning AI physics models.