AI Developer — Montréal

Maxime Gagné
Building AI systems that think.

Self-taught developer specializing in multi-agent architectures, local LLM systems, and AI-powered automation.

Each project below is a working system — not a demo or a tutorial follow-along. Click a project to learn what it does, then visit the GitHub repository for the full technical details.

LinkedIn ↗ maximegagne.ai@gmail.com GitHub ↗

Projects

🧠

Local Agentic System

SecondMind

An AI assistant that runs entirely on my own computer — no cloud, no subscriptions. 9 specialized AI agents collaborate to understand context, remember past conversations, audit their own code, and improve over time.

See this project →
⚙️

SaaS Concept

AI Workflow Discovery Agent

A hybrid system that clusters raw business data with HDBSCAN, then runs a 3-agent LLM pipeline to diagnose friction, design an optimized workflow, and export a deployment-ready Apache Airflow DAG — with ROI calculated deterministically before any LLM call.

See this project →
☁️

Cloud / Azure

Azure Auditeur RAMQ

A cloud pipeline built on Microsoft Azure that reads handwritten medical notes and extracts billing information automatically — designed for the Quebec healthcare system (RAMQ).

See this project →

Browser Architecture / LLM

Prompt-Pad: Logic Sampler

A zero-telemetry injection system built on an isolated Shadow DOM. It solves orchestration friction across ChatGPT, Claude, and Gemini using Manifest V3 and Service Workers to bypass strict CSPs while guaranteeing 100% local data privacy.

View Technical Whitepaper →