Lucas Maingi

I’m an AI/ML and full-stack engineer focused on designing, building, and deploying end-to-end intelligent systems—from predictive ML models to agentic LLM architectures.

I build clean, responsive interfaces with Next.js/React, and I orchestrate high-performance, asynchronous backend APIs using Python with FastAPI/Flask. I also develop robust data ingestion pipelines and real-time integrations, including payment gateway work with Safaricom M-Pesa Daraja API.

I care deeply about engineering excellence and reliability: I use Docker for containerized environments, Pytest/Playwright for unit and integration testing, and GitHub Actions for automated CI/CD pipelines. In LLM/agent systems, I prioritize governance, safety, and cost controls through layered execution, verification checks, and secure API gateway patterns.

My standout work includes Hermes (WhatsApp Commerce + M-Pesa flow with a webhook-driven payment lifecycle and an 83-test suite), and LLM-driven systems like SentryBrain (autonomous remediation with manual validation gates) and Argus (CI/CD regression gating for agent performance). I’m also building explainable fraud detection with hybrid models and actionable SHAP insights.