Teissier YannisDeveloper · AI & fullstackLyon · Sept. 2026

I build software for the web, and sometimes for machines that learn.

that's the whole pitch

CurrentlyAI software developer at Hozzway, Lyon. Fullstack Python & React.

Working onPutting AI systems into people's hands: proofs of concept, integration, backend and DevOps around the models.

ElsewhereGitHub · LinkedIn · Uses

Selected work

6 projects, 2024 – 2026§ 01
  1. 01Contre-FrappeClient, authoritative dedicated server, bots, GOTV, map editor and launcher, one engine codebase, still in progress.Rust · Bevy · lightyear · Tauri · Fastify · KubernetesRust, Bevy, lightyear, Tauri, Fastify, Kubernetes2024–26
  2. 02Homelab GitOpsK3s across a rented VPS and a machine at home, Argo CD app-of-apps, one Helm chart per app, Jenkins bumping tags, sealed secrets. It deploys everything else on this page.K3s · Argo CD · Helm · Jenkins · Traefik · WireGuard · Sealed SecretsK3s, Argo CD, Helm, Jenkins, Traefik, WireGuard, Sealed Secrets2026
  3. 03AltiTrakGPX and Strava activities turned into vertical speed, climb splits, run/hike ratio, training load and a "you vs the race" readiness score. FastAPI, PostGIS and TimescaleDB, React.FastAPI · PostgreSQL · PostGIS · TimescaleDB · React · PlaywrightFastAPI, PostgreSQL, PostGIS, TimescaleDB, React, Playwright2026
  4. 04A modded Minecraft server, in production545 mods, a handful of players, and a tick that went from 120 to 45 ms, every change backed by a profile, every fix a twenty-line mixin, and a supervision panel that cannot write to the server by construction.Go · Java · NeoForge · Spark · Fastify · React · PostgreSQLGo, Java, NeoForge, Spark, Fastify, React, PostgreSQL2026
  5. 05Real-time vision on the edgeTwo TensorRT detectors in series, a tracker that keeps the blur stable, zone counting and a heatmap, all in one GPU pipeline with a live web view, developed on a desktop GPU, written to be carried to a Jetson Orin Nano.DeepStream 7.1 · TensorRT · GStreamer · Python · ONNXDeepStream 7.1, TensorRT, GStreamer, Python, ONNX2026
  6. 06RAG chat, multi-cloudDocument ingestion, vector search, reranking and streamed answers with sources, every model call going through one LiteLLM proxy that fails over between Bedrock, Azure AI Foundry and Vertex AI, or runs fully local.FastAPI · React · LiteLLM · Qdrant · PostgreSQL · Redis · TerraformFastAPI, React, LiteLLM, Qdrant, PostgreSQL, Redis, Terraform2026
Fig. 1 — the pieces · launcher, client, dedicated server with its bots, backend, GOTV relay01
Fig. 2 — the cluster · a rented VPS holds the control plane, a home node runs every workload, WireGuard between them02
Fig. 3 — from a track file to a readiness score · and the climbs the app actually cares about03
Fig. 4 — above, the supervision chain and the wall the collector cannot cross; below, five bottlenecks before and after04
Fig. 5 — one GStreamer pipeline on one GPU · decode, two detectors, tracker, a Python probe that pixelates and counts, MJPEG out05
Fig. 6 — ingestion on top, a question below · one proxy in front of three providers, and the answer streamed with its sources06
no screenshots, drawings

About

§ 02

i, I'm Yannis. I live in Lyon and work as an AI software developer at Hozzway: fullstack Python and React, and the work of turning AI models into systems people actually use. Before that, I rebuilt an embedded network appliance at Sheltercom, did data science and backend work at Sogeti, and graduated as an engineer from CPE Lyon. If you're curious about the tools I use day to day, there's a uses page for that.

Off the clock I build things that need to run for real: a game server, a Kubernetes cluster at home, a trail-running app for my own races. I also run and ride, travel and take photographs. Always happy to hear about a new idea: drop me a line.

“From a Rust game server to a GitOps cluster, one measurement at a time.”