ENFR
Portrait of Jescy Gratton

About

Jescy Gratton

Self-taught developer based in Repentigny, Quebec.
Building sober, compliant, and useful products — for organizations that don't have Google's resources.

  • Repentigny, QC
  • 6+ years in code
  • Loi 25 / Loi 96 by construction

I’m Jescy, a self-taught developer based in Repentigny, Quebec, Canada. My path isn’t conventional: no computer science degree, but more than 6 years of coding and an persistent obsession with systems that actually work in the hands of real people.

I built my expertise the way you learn a trade — by doing. The research I conduct in parallel on neural architectures remains unpublished to date: I prefer to validate before sharing.

What drives me

Who I work for

Community and public sector — nonprofit organizations (OSBL), public bodies, cooperatives, mutual associations. Because the mission matters. I often start collaborations on a pro bono basis: it lets us verify fit, deliver real value, and build a relationship that can evolve into paid work.

Quebec tech SMBs — for whom applied AI (RAG, classification, automation) or professional web infrastructure provides a concrete operational advantage.

Augmented AI — a skill, not a shortcut

I use daily Claude (Anthropic), GPT-4 (OpenAI), local Ollama models (Llama 3.1, Qwen 2.5) and the LangChain / Pinecone / Qdrant stack in production. Not as a substitute for competence, but as a multiplier: I deliver 3–5× faster with superior quality because I know which tool to use for which task, how to direct it, how to verify its output, and where its limits are.

This is a skill in its own right in 2026. I built it through practice on my own neural architecture research projects before transposing it into client deliverables.

What this means for you: a realistic delivery timeline that was 3 weeks becomes one week. Not by sacrificing quality — on the contrary: with an additional layer of automated review (Loi 25/96 audits, accessibility, security) that few developers can afford.

Stack

Languages — TypeScript, Python, Rust (active learning), Bash

AI / Machine Learning — PyTorch, Hugging Face Transformers, LangChain, scikit-learn, custom neural architectures

Web / Backend — Astro, Next.js, FastAPI, Node.js, PostgreSQL, Redis

Infrastructure — Docker, Cloudflare Workers, Vercel, AWS (working knowledge), Linux

Tools — Git, Neovim, Figma (reading)

AI in production — Claude (Anthropic), GPT-4 (OpenAI), Llama 3.1 / Qwen 2.5 via Ollama (local-first), LangChain, Hugging Face Transformers, RAG (Qdrant, pgvector), agents

Living list: what I currently use in production or R&D. No percentages — experience is measured in delivered projects, not progress bars.

Quebec Compliance

Every site I deliver respects Loi 25 (Quebec privacy law) and Loi 96 (Quebec French language charter) by construction — verified at every commit via an automated audit I built for my own projects (see loi25.jescygratton.ca).