ACHILLE TRIOMPHE

About — Prosopon

The face one turns to the world.

Hey! I'm Achille. I work on reinforcement learning and hardware-aware development. I believe we've finally reached the promised land of digitalization, data, and personalisation. RL is at the center of this revolution: it is what makes personalization at scale possible through turning real processes into environments we can train against.

I started computer science at EPFL in Lausanne, Switzerland. I learned so much from great professors and passionate students with access to state-of-the-art infrastructure. I'm grateful for my friends and the people I met there. After five years at EPFL, I interned for eight months at Sinequa (ChapsVision) in Paris, where I worked on evals for agentic RAG systems. I compared LLM-as-a-judge and deterministic setups, and designed an automated pipeline for retrieval parameter optimization. Since then, I've decided to pursue my own study program to specialise in RL and hardware-aware development. I want to build things from scratch and take advantage of the freedom I have right now to work with the tools that will shape the coming years.

A little context on who I am, for the curious souls. I'm French, but I've lived most of my life across Tokyo, Singapore, Hong Kong, and Lausanne. I've loved every moment of discovering these cultures, the languages, and meeting friends from all over the world. Sport is a central part of my life. I've pursued a large variety at an amateur level (football, tennis, track and field) and at a competitive level (swimming and rock climbing). Since moving to Switzerland, I've pivoted to alpine sports: mountaineering and trail running.

Where I'm at

Block 1 Foundations

Linear Algebra & Numerical Computing for AI

  • Matrix decompositions with numerical stability
  • Automatic differentiation from first principles
  • GPU-accelerated matrix operations (CUDA / Triton)

Experience

  • Jan — Aug 2025 Paris, France

    AI Engineer Intern

    Sinequa

    Context engineering and evaluation for Agentic RAG — built the evaluation backbone that made retrieval systems measurable, tunable, and reliable for client deployments.

    • Pioneered a no-LLM evaluation framework using annotated synthetic data, improving internal benchmarks and industry-standard evals (BeerQA, Multihop-RAG) by up to 0.2
    • Established the groundwork for automated system validation and hyperparameter tuning for client deployments
    • Systematically analyzed neural and hybrid search parameters to inform key architectural decisions on retrieval accuracy
    • Extended the evaluation pipeline to multi-turn, conversational Agentic RAG
    • Contributed to a major refactoring of the internal evaluation library, reducing technical debt and improving maintainability
  • Oct 2024 Zurich, Switzerland

    Case Study Winner

    ETH Hackathon

    Built an intelligent flashcard generator that turns any notes into interactive, visual learning material — out of 100+ participants and 20+ projects.

    • Pipeline: Tesseract OCR for text extraction, spaCy NLP for key-concept identification, OpenCV for visual detection
    • Multi-modal analysis with Claude Sonnet to generate text-integrated visual flashcards
  • May 2021 — Jul 2024 Lausanne, Switzerland

    President

    START Lausanne

    Led 25–30 active members (and a 150+ alumni network) running Europe's largest student-led early-stage startup competition.

    • Record prize of CHF 35,000 — the largest in START's history
    • Mentored 60 projects impacting over 120 students
    • Directed hackathons, startup weekends, pitch competitions, and VC meetups across Lausanne and Europe
  • Dec 2023 Lausanne, Switzerland

    Winner

    EPFL Hackathon

    Developed an LLM-powered polling app to help students debate and plan the university's sustainable future — out of 300+ participants and 100+ projects.

    • Presented findings to EPFL's sustainability and innovation committees
  • Jan — Apr 2023 Paris, France

    Technology Consultant & Developer

    Hussar Academy

    Shipped a no-code-friendly workflow automation and database management system, improving operational efficiency.

    • Led the platform from conception to launch: market research, competitor analysis, user testing

Education

  • Sep 2020 — Feb 2025 Lausanne, Switzerland

    École Polytechnique Fédérale de Lausanne

    B.Sc. Computer Science
    • Systems & hardware: OS internals in C (CS-323), computer architecture (CS-208/209), concurrent programming in Scala (CS-206), multiprocessor architecture with OpenMP, pthreads, and CUDA (CS-307), data-intensive systems in C and SQL (CS-300)
    • Software engineering: Android app in Kotlin with Firebase, CI/CD, testing (80%+ coverage), agile team workflows (CS-311); functional programming in Scala (CS-214); system programming in C with POSIX, OpenSSL, and client-server (CS-207/212)
    • Compilers & theory: Computer language processing in Scala — lexer, parser, type checker, code generator (CS-320); theory of computation (CS-251), computer security and privacy (COM-301)
    • ML & data: Numerical methods for ML with PyTorch and automatic differentiation (CS-328), internet analytics with Hadoop and Spark (COM-308), signal processing (COM-202), stochastic models (COM-300), probability & statistics (MATH-232)

Achievements

Academic

  • CS-311 The Software Enterprise — Teaching Staff's prize for best project (Dec 2024)
  • EPFL Hackathon winner (2023)

Sports

  • Trail running — Restonica 100k: 70km, 3900m D+ (2026)
  • Trail running — Transylvania Trail 70k: 50km, 3300m D+ (2025)
  • Mountaineering — Mont Blanc ascent (2023)
  • Rock climbing — 3rd in Paris Championship (2018)
  • Swimming — 100m backstroke: ranked #1 in France in my age category (2016)

Languages

French Native
English Cambridge C2
Spanish DELE B2
Mandarin Chinese HSK 4 — B2

Contact

GitHub Twitter LinkedIn Email