Skip to content
View SylvainVerdy's full-sized avatar
🎯
Focusing
🎯
Focusing

Highlights

  • Pro

Block or report SylvainVerdy

Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
SylvainVerdy/README.md

header

Hits

🚀 AI/MLOps Engineer building IAFactory — an internal LLM inference platform on Kubernetes

🔬 Ex-NLP Research Engineer · Published at TALN 2023 & KES 2024 · CIFRE PhD (NXP/LIA)

🎓 Engineering degree from ESIEA Paris (highest honors)

"Some people build models. Others build the infrastructure to run them. I do both."


🏗️ What I'm working on

class AIFactory:
    """Internal LLM Inference Platform"""

    stack = {
        "inference":    "vLLM or SGLang?",
        "gateway":      "Istio? Envoy Gateway? Apisix?, what else ?",
        "scheduling":   "Mystery",
        "deployment":   "(GitOps)",
        "observability": "Grafana · Prometheus · Loki · DCGM Exporter· OpenTelemetry",
    }
    current_model = "your_open_source_model <3"

    def iterate(self):
        while True:
            self.optimize_inference()
            self.harden_quotas()
            self.ship()

⚡ Tech Stack

Python Kubernetes Docker Helm ArgoCD Grafana Prometheus NVIDIA

PyTorch HuggingFace TypeScript React Lua Spark


📊 GitHub Stats

SylvainVerdy's github stats Top Languages Card


📌 Featured Repos

rag_kube LLM-FT-RELATION-EXTRACTION DWIE-FR end-to-end-SLU


trophy

footer

Pinned Loading

  1. rag_kube rag_kube Public

    Système RAG (Retrieval-Augmented Generation) pour interroger vos documents via une interface web, avec déploiement Kubernetes, observabilité (Langfuse, MLflow) et CI/CD.

    Python

  2. ItinaryMe ItinaryMe Public

    TypeScript

  3. Agent-S Agent-S Public

    Forked from simular-ai/Agent-S

    Agent S: an open agentic framework that uses computers like a human

    Python

  4. Emvista/DWIE-FR Emvista/DWIE-FR Public

    DWIE-FR : A new French dataset annotated with named entities

    Python 4

  5. end-to-end-SLU end-to-end-SLU Public

    Forked from lorenlugosch/end-to-end-SLU

    PyTorch code for end-to-end spoken language understanding (SLU) with ASR-based transfer learning

    Python

  6. sauerkraut-lfm2-gliner sauerkraut-lfm2-gliner Public

    A from-scratch reproduction of the build pipeline behind VAGOsolutions/SauerkrautLM-LFM2.5-GLiNER: a zero-shot, multilingual Named-Entity-Recognition model that turns the causal LFM2.5-350M hybrid …

    Python