Studio › Candidates › Yohann Lesueur NEW UPLOAD Contact
PROCESSING COMPLETE A new creator just uploaded himself to your candidate pipeline.

Senior Frontend Engineer,
ready to publish.

I'm Yohann Lesueur — React & TypeScript engineer, seven years shipping to production. For the last two, I built exactly what Studio is: a data-heavy dashboard where non-technical users control a complex system — real-time sensor fleets instead of video libraries, same battles: architecture, performance, clarity. Scrub through the story.

CHAPTER 1 — 2016

École 42, the C piscine

One month of C, git and shell. The month that decided everything.

2016 / 2026

Content — my library, all shipped to production

The dashboard rebuild: legacy jQuery → React/TS clean architecture

ACCENTA · ENERGY PLATFORM · REACT 18 · TYPESCRIPT 5 · REDUX TOOLKIT

A real 445-line-component codebase, sparse docs, zero tests on the view layer. I introduced a framework-agnostic core/: domain models (DDD bounded contexts), ports & adapters, use-cases in TDD, dependency injection down to the date provider. The view became a thin presenter over Recharts. Exactly the “explore unfamiliar codebases, turn messy problems into clean solutions” your posting describes.

Clean architectureDDDTDD · Jest/RTLRedux listener middleware

Time-series dataviz that survives the year view

SENSOR FLEETS · RECHARTS · MONGODB AGGREGATION · PERFORMANCE

Sensor fleets producing readings around the clock, and users asking for 12-month views. Strategy: never render the raw stream — server-side aggregation by user-chosen granularity (minute/hour/day pipelines), adaptive Line↔Bar rendering, dot=false, client-side zoom with no refetch, memoized derived series. Performance as an architecture decision, not a patch.

Web performanceRechartsUX under load

Agentive: an AI-agent platform, built and operated solo

IDEM AGENCY · PYTHON/FASTAPI + REACT · LLM · CI/CD

Co-founded a two-person studio — no ops team, my own services in production. Multi-provider LLM abstraction with fallback, cost estimation and secret masking; pgvector RAG; sandboxed tool execution; architecture contracts enforced in CI. Fullstack autonomy, by necessity.

AutonomyAI in productionCI/CD

Loqua: a privacy-first voice coach that measures itself

SOLO · LLM INTEGRATION · EVAL HARNESS · GDPR

Structured outputs, strict schema validation, typed error handling — and an evaluation harness of 60 reference cases scored by an LLM judge, so model upgrades are measured, never guessed.

LLM evalsProduct sense

Analytics — the numbers behind the candidate

0years shipping to production
0frontend paradigms migrated
(jQuery legacy → React/TS)
0reference cases in my LLM
evaluation harness
0% of this site hand-reviewed —
built with AI, owned by me
Skill engagement — retention over a decade React / TypeScript Python / APIs

Job fit — your posting, line by line

  • React & TypeScript — “your everyday tools”Daily since 2021. React 18 + TS 5 SPA in production; typed domain models down to sensor payloads.✓
  • Complex dashboards & interactive UXTwo years on creator-grade dashboards: filters, thresholds, multi-format exports (CSV/XLS/PDF/PNG), i18n, multi-tenant theming.✓
  • Web performance fundamentalsServer-side aggregation strategy, adaptive chart rendering, memoization, zoom without refetch, Webpack 5 builds.✓
  • Testing, docs, best practicesTDD use-cases, Jest + RTL, success/failing stubs, object-mother factories, a dedicated test harness. Specs written before code.✓
  • Navigate undocumented codebasesMy defining project: a sparse-doc legacy front, explored, mapped, and rebuilt into clean architecture — while the product kept shipping.✓
  • GraphQLHonest gap: my production data layers are REST/DRF. The hard parts — fetching, caching, normalizing client-side state — are my daily work; the query language is a fast bridge, and I'm already on it.→
  • AI tools, with known limitsI run Claude Code daily and review every generated line like it's mine. This site is the live demo: AI-built, human-directed, three critique passes.✓
  • English as working languageProfessional level (C1) — this entire candidacy is the writing sample.✓

Monetization — what Studio gets from day one

SHIP

A senior who has owned features from spec to production — and kept them alive after. No hand-holding required; updates shared before you ask.

HARDEN

A test culture that sticks: use-cases in TDD, error paths tested explicitly, harnesses your teammates will actually enjoy building on.

CLARIFY

Someone who has turned a 557-line component soup into bounded contexts — and can do the same for any corner of Studio that grew too fast.

CARE

Creator-first instincts: I spent two years making complex data legible for non-technical users. Your creators' dashboards are my home turf.

Ready when you are, Marvin.

30 minutes to see if this fits? I'll bring honest answers about the GraphQL gap and strong opinions about creator dashboards.

lesueur.yohann@hotmail.fr github.com/yoles LinkedIn

French native · English professional (C1) · Rennes, FR — ready for the Paris hybrid rhythm