Judgment first
Architecture, data model and trade-offs are decided before any agent writes a line. Agents follow a plan; they don't make it.
Open to a full-time role · Paris area
Juan Restrepo
Full-stack developer building with AI agents
I've been building professional apps with React and Python for 3 years, and I run my whole development process with AI agents: specs, implementation, tests and code review.
About
Full-stack engineer, three years in production
Full-stack engineer with 3 years of experience shipping apps in production. What I enjoy most is turning an idea into something real: I design it, build it end to end, put it in people's hands and keep improving it.
I treat AI agents as part of the team. I write the spec first, then orchestrate agents that plan, implement, test and review the code, and I stay accountable for every line that ships.
- Location
- 92600 Asnières-sur-Seine, France
- Education
- Master's in Engineering · ENSEA
- Languages
- Spanish (native) · French, English, Portuguese (C1) · Italian (B1)
- Looking for
- Full-time role (CDI) · Paris area, open to relocation
- Work authorization
- Authorized to work in France · no sponsorship needed
AI workflow
AI in every step, judgment in every step
I use AI agents to build faster, never to think less. They take part in every step, and I stay responsible for what ships.
Design
I decide the architecture and the trade-offs; AI helps me explore options.
Plan
The work is split into small tasks, each with a clear way to check it.
Build
Agents write the code and the tests, several tasks in parallel.
Review
Reviewer agents and I check every change before it moves on.
Ship
Automated tests run and the app deploys, with nothing left to manual steps.
Every change reviewed
I read the code agents write the way I'd read a teammate's, and automated tests back every change. If I can't explain it, it doesn't ship.
AI inside the product
I also build AI features into apps: agents that use the app's own data as tools, with guardrails so they can't invent answers.
Team adoption
I brought Claude Code and MCP into my team's daily work: shared project rules and automated code review.
Experience
Where I have worked
Full-stack developer · AI-driven development Current
Jan 2026 – nowOlenergies ↗ · Bagnolet, France
Olenergies is a French energy startup that designs and builds batteries and mobile power units: a silent, cleaner alternative to diesel generators. It sells to businesses, around ten clients, who use the units at events and on construction sites, or rent them out.
- Built on my own the two apps that run the installations, working with AI agents: the web platform where clients follow each installation live (map, alarms, consumption forecasts) and the touchscreen that controls it on site.
- Brought AI agents into the team's daily work (Claude Code, MCP), with shared project rules and automated code review on every change.
- Designed a multi-site architecture where each client only sees their own installations, and a secure gateway to reach them remotely without exposing them online.
- Set up CI/CD pipelines: every change is tested and deployed with zero downtime.
Full-stack and IoT developer (work-study)
2023 – 2025Olenergies ↗ · Montreuil, France
- Built React dashboards showing solar production, battery charge and alarms in real time.
- Developed the backend that collects device data and serves it live to the dashboards (REST API, WebSocket).
- Took part in the whole product cycle: specs, tests on benches and real sites, deployment and maintenance.
Stack
Tools I use every day
AI and agents
Frontend
Backend
Infra and cloud
Quality and CI/CD
Projects
Selected work
OlenPeps
Lets businesses follow their solar and battery installations from anywhere: battery level, solar production and alarms, live on a map. They can even open an installation's own touchscreen from the browser, as if they were standing next to it.
Try the demo ↗
Private code
MoviesApp
Helps you decide what to watch tonight. Browse movies and shows, see where they stream in your country, or just tell the AI assistant what you feel like and get a few picks, each with a reason.
Monthwise
Shows where your money goes each month. Add an expense in a few taps, see your balance and spending by category, and compare with past months to know if you are spending more than usual.
Steam Backlog Coach
Tells you what to play today. It looks at your Steam library, shows what you have never started or left halfway, and suggests a game for the time you have and the mood you are in.
Contact
Let's talk
Have a role or a project in mind? Write to me.