I turn messy workflows into useful software, then make sure it holds up.
Greater Boston · Haitian roots 🇭🇹 · English / Français / Kreyòl
I'm a software and reliability engineer focused on AI solutions and implementation. My background spans 10+ years in technology, from enterprise applications and customer support to cloud systems, APIs, and production reliability.
These days, I build AI applications and automation that solve concrete problems. My SRE experience at Liberty Mutual shapes how I work: understand the failure modes, make the system observable, and leave the next person a useful runbook.
- AI that does useful work: tool calling, structured outputs, and workflow integrations.
- Software people can depend on: clear APIs, tests, error handling, and CI/CD.
- Technical depth with business context: requirements discovery, troubleshooting, and explaining trade-offs to the people using the system.
- A legal-tech perspective: 4+ years across two law firms, working with enterprise applications and the workflows behind them.
A good demo gets attention. A dependable system earns trust.
|
A conversation with my work. An AI assistant grounded in my professional background. It answers questions, captures visitor contact details, and records questions it cannot answer. Built with bounded tool calls, rate limiting, structured logging, and graceful provider-error handling.
|
Your statements. Your machine. A local-first desktop app that turns PDF bank statements into cash-flow, spending, and recurring-charge insights. Deterministic code computes the numbers. An optional AI layer explains sanitized results after a privacy gateway strips personal information.
|
More projects & engineering stories ↗
Python TypeScript React Next.js FastAPI Node.js
| Where I work | What I use |
|---|---|
| AI applications | OpenAI API, Gemini, tool calling, Python |
| Web & mobile | React, Next.js, React Native / Expo, Tailwind CSS |
| APIs & data | FastAPI, Node.js / Express, PostgreSQL, Supabase, SQLite |
| Cloud & delivery | AWS Lambda, API Gateway, CDK, DynamoDB, GitHub Actions |
| Reliability & testing | Datadog, CloudWatch, BigPanda, pytest, Ruff |
| AI development tools | Claude Code, Codex |
Understand the workflow → Build the smallest useful system → Verify it
↑ ↓
Learn from real use ← Ship & observe
Make AI output checkable. Keep deterministic logic where correctness matters. Validate model output before it drives a workflow.
Plan for failure. Timeouts, bounded retries, useful logs, and understandable fallback messages belong in the design.
Keep the human in view. The person using the system should understand what happened and what to do next.
A little more about me
- B.S. in Computer Science from Boston University.
- Hack.Diversity Software Engineer Fellow alumnus.
- Originally from Haiti, now based in Greater Boston.
- I speak English, French, and Haitian Creole.
- Exploring retrieval-augmented generation, agents, and MCP through hands-on builds.
- Most of my project ideas start with: “Why is this process still so painful?”
I'm open to AI solutions, implementation, software, and reliability engineering opportunities.
Greater Boston or US remote.
Let's talk / See my work / Connect on LinkedIn
Build useful. Ship reliable.