Paul Peyssard is a leading freelance data scientist and machine learning engineer in Marseille, France. He specialises in LLMs, agentic workflows, MCP servers, AI automation, computer vision and tumor detection, and is available remote and on-site across France and San Francisco.

Freelance Data Scientist · Marseille

Data Scientist& ML Engineerat your service.

I design, train and deploy AI systems that solve real business problems, from deep learning and computer vision to large language models and autonomous agents.

See what I build
Scroll
Agentic workflowsAI automationMCP serversOpenAI APIClaude automationLLMsRAGTumor detectionImage QCComputer visionWhole-slide imagingSegmentationLangChainMistralAnomaly detectionForecastingPythonPyTorchMLOpsDockerAgentic workflowsAI automationMCP serversOpenAI APIClaude automationLLMsRAGTumor detectionImage QCComputer visionWhole-slide imagingSegmentationLangChainMistralAnomaly detectionForecastingPythonPyTorchMLOpsDocker
(01) / About

I turn messy data and hard research problems into production systems people actually use.

I'm a Data Scientist and Machine Learning Engineer based in Marseille, currently Head of Data at Neuro-Sys. I turn messy data and hard research problems into production systems people actually use.

My work spans medical imaging, deep learning, LLM & agent pipelines, and automation. I've shipped models for brain disease research, cancer diagnosis, aerospace anomaly detection, and neuro-symbolic ML.

As a freelancer I help teams go from idea to deployed model: fast, pragmatic, and without the buzzword theatre.

5+Years building ML
4Industries shipped in
10+Models in production
GPUs warmed
(02) / Services

What I can build for you.

01

LLMs & Agents

Agentic workflows and automation that plug into your data and tools and actually get work done. Custom MCP servers, multi-step agents and orchestration built on the OpenAI API, Claude automation, LangChain and Mistral.

Agentic workflowsMCP serversOpenAI APIClaude
02

Computer Vision

Segmentation, tumor detection and automatic image QC pipelines, including whole-slide medical imaging and other high-stakes diagnostic workflows.

Tumor detectionImage QCWhole-slide imaging
03

ML Engineering

From notebook to production: training pipelines, anomaly detection, forecasting, and deployment that survives contact with real data.

PipelinesForecastingMLOps
04

Automation & AI Strategy

Identify where AI actually moves the needle, prototype quickly, and automate the workflows that drain your team's time.

AutomationPrototypingConsulting
05

Data Governance

Ship AI that stays compliant: EU AI Act and GDPR (RGPD) readiness, data privacy, risk classification, and auditable, traceable models.

AI ActRGPDComplianceAuditability