Sep 2025 – Nov 2025 · 2 mos

• building a general MacOS AI Agent, backed by a16z speedrun
• I joined pre-product, and built the entire LLM agentic system, including:
• in-house browser use (did beat browser-use.com accuracy on internal testing! thanks to a novel technique I developed leveraging code generation)
• computer use (CUA)
• MCPs clients and servers
• memory / retrieval systems
• also mentored 2 junior engineers! :)

Apr 2025 – Sep 2025 · 5 mos

• building an AI UGC Ads generation product, $1M annual revenue per employee!
• I built:
• emotion control for Arcads AI actors -> led to Arcads most viral feature launch of 2025 (on linkedin)
• v1 of the video editing LLM agent
• the entire data infrastructure of the company
• I also started building the AI team (by sourcing, interviewing, and hiring our 2nd AI Engineer)! :)

Apr 2024 – Apr 2025 · 1 yr · San Francisco, CA

• building an entirely autonomous google ads agency, backed by Entrepreneur First
• I worked on:
• finetuning an LLM on tens of thousands of scraped quality website designs (-> surpassed existing LLMs design quality by a lot, near 100% preference rate)
• training a google ads ai agent to maximize Ads ROI (with a novel method to efficiently teach textual human feedback to the model)
• some more facts:
• managed $1M annualized ad spend (entirely automatically! with +2x average ROI (not ROAS))
• backed by Entrepreneur First & Transpose Platform
• acquired by Arcads.ai (one of the best AI Ads companies in the market | $1M rev / employee)
• shout out to Lucas Marandat, my amazing cofounder now at Mistral ; and to all of our team: Lanzelot Moll, Nicolas Bellot, Gildas Garin-Hameline!

Sep 2023 – Apr 2024 · 7 mos · San Francisco, California, United States

• Backed by top founders including Demis Hassabis (DeepMind + Nobel Prize Winner), Alexandr Wang (Scale AI), Patrick and John Collison (Stripe), Reid Hoffman (LinkedIn), Tom Blomfield (Monzo), Nat Friedman.

Sep 2021 – Jun 2023 · 1 yr 9 mos · Toulouse, Occitanie, France

• Reinforcement learning + continual learning (model training & evaluation). Applied to the space/satellite industry (for Thalès/Airbus).

Oct 2020 – Aug 2021 · 10 mos

• Leader of a finalist team at the NeurIPS 2020 MineRL competition
• Contributor on Neural MMO: OpenAI's first Massively Multiagent Environment for RL Research

Jun 2019 – Aug 2021 · 2 yrs 2 mos

• Developed internal software products for major french companies (+100k employees).

Jun 2017 – Jun 2017 · 0 mo

• Cryptography, hardware security and software obfuscation internship.
• Was invited to the annual conference of the company, meeting international partners and attending technical talks.

Oct 2015 – Oct 2015 · 0 mo

• Pentesting & web dev internship.