web
Static HTML/CSS via Jekyll, GitHub Pages native build (no framework, no JS toolchain, no CI to write — GitHub Pages builds Jekyll automatically). Decided during design brainstorm, recorded here rather than asked again.
Recruiters and hiring managers at French AI/ML companies evaluating Arnaud Thery for AI/ML engineer roles, arriving via a link shared in job applications or on LinkedIn. Secondary: Arnaud himself, sharing the link as his public professional presence.
A public complement to the PDF CV generated by Diggo (his job-search pipeline) — a fast, always-available overview for a recruiter who won't open a PDF, plus a portfolio section with more project depth than a CV allows. Success: a recruiter gets a credible, scannable impression in under a minute and can go deeper (CV, projects, GitHub) if interested.
Senior on relational maturity, still building on the technical stack: 8 years managing a team (Fnac, department manager) — communication, prioritization, pressure management — combined with 2+ years of hands-on production AI/ML engineering (Missia). The site should give the management experience real weight, not bury it under a skills-tag list, and pair it with concrete technical proof (shipped systems, not just tools used).
Shared as a link in job applications and on LinkedIn, alongside the Diggo-generated PDF CV and cover letter — not a replacement for either. Recruiters skim in seconds before deciding whether to open the CV page or click through to a project's GitHub repo.
- Static content only — no backend, no contact form with server-side processing, no CMS.
- Bilingual FR/EN, both languages fully written (not a translated
afterthought) — content rewritten once for the web from
diggo/config/cv.yaml, then maintained independently (no sync pipeline). - No custom domain for v1 (
github.iosubdomain).
Name: Arnaud Thery. No existing personal logo, visual identity, or stated voice — first personal site.
Real, checkable content only (source: diggo/config/cv.yaml):
- Experience: AI/ML Engineer at Missia (Sept. 2023–present, production FastAPI services, AWS SageMaker, edge inference on Jetson Orin NX); AI Developer internship at GoodBarber (semantic search engine); Fnac department manager (2015–2023, 8 years, team of 8).
- Projects with public repos: diggo, rag-lab, Kaggle Watson, Restaurant Analytics (live demo on Railway).
- Two merged open-source contributions: mjpost/sacrebleu, castorini/pyserini.
- Certifications: Claude Code 101 and Claude 101 (Anthropic, 2026), PSM I (Scrum.org, 2026), Google Project Management, CI/CD for ML, LLM/MCP Fundamentals (Hugging Face), Google Cybersecurity, others (2025).
- Education: MSc in AI (Aflokkat/ESTIA, 2024–2026) plus prior RNCP certifications in AI/software development.
No testimonials, press, or third-party endorsements exist — none to be invented.
- Credibility over hype: every claim backed by something checkable (a GitHub link, a live demo, a merged PR) — no unverifiable superlatives.
- The management years are a feature, not a footnote: give them real narrative space, not a buried bullet under a technical skills list.
- Scannable first, detailed second: the homepage must land the value proposition before any scroll; depth lives on the CV/Projects pages.
- Bilingual parity: FR and EN pages are equally complete.
- Low maintenance: fully static, no infrastructure to babysit during an active job search.