You're seeing this page as if you were . The main menu is still yours, though. Exit from immersion
Jérôme DragoJD

Jérôme Drago

Supermalter

Senior AI Engineer | Agentic AI, Multi-Agents

750 €/jour
11 projets
Paris, FR
8-15 ans

Délai de réponse moyen : 1h

À propos de Jérôme

With over ten years of experience, I specialize in developing advanced agentic AI systems and cognitive simulations at Sanofi. My work focuses on leveraging tools like AWS Bedrock AgentCore, Strands SDK, and Retrieval-Augmented Generation to build production-grade systems that integrate psychological principles into AI design. My initiatives include prototyping and deploying agentic cognitive simulation platforms and designing scalable architectures compliant with pharma industry regulations.

By combining a background in clinical psychology with technical expertise, I apply validated psychological frameworks such as Big Five/OCEAN and Atkinson-Shiffrin memory models to agent design. This approach ensures consistency, memory retention, and character fidelity in AI systems. My commitment lies in creating innovative solutions that drive impactful and regulated advancements in AI technologies.
  • Français

    Bilingue ou natif

  • Anglais

    Capacité professionnelle complète

Accepte de travailler sur site
Paris (jusqu’à 10 km)

Expériences

  • Sanofi - Digital Accelerators
    Logo MaltSur Malt
    Sanofi · Senior Software Engineer, Agentic AI & Cognitive Simulation
    INDUSTRIE PHARMACEUTIQUE
    janvier 2026 - Aujourd'hui (7 mois)
    Paris, France
    Originated, prototyped, and led the technical development of an agentic cognitive simulation platform — from solo POC to cross-disciplinary production system, presented at VivaTech 2026.
    Architected the full prototype solo: Python/FastAPI backend on AWS Lambda, TinyTroupe integration with custom Bedrock adapter, OpenSearch Serverless semantic memory, SSE streaming via Lambda Web Adapter, React frontend — deployed on AWS with SST

    Migrated to AWS AgentCore v2: designed the cognitive turn core using Strands Agents SDK, AG-UI native SSE protocol, Jinja2 prompt assembly with Bedrock prompt caching (~90% token savings on static identity block), and Pydantic structured output (TALK/THINK/DONE)

    Wrote 17 Architecture Decision Records covering memory isolation, cognitive state persistence in agent.state, single-shot structured output, and multi-tenant brand scoping — establishing the technical foundation for the production v2 system

    Bootstrapped Sanofi Accelerator's Claude Code + Cursor agent/skill framework (ai-standards library) adopted across multiple teams
    Bedrock agentcore strands Agentic AI Terraform
  • Sanofi
    Sanofi · Senior Software Engineer, Launch Intelligence Platform
    INDUSTRIE PHARMACEUTIQUE
    mars 2025 - janvier 2026 (10 mois)
    Paris, France
    Contributed to SILC, an internal platform for tracking pharma product launch metrics across Sanofi.
    Built a dual-runtime data ingestion pipeline (V1/V2 Lambda router with LaunchDarkly feature flag, S3 event parsing, date metadata normalization for Pinecone KB filtering) enabling zero-downtime migration between knowledge base versions
    Developed the launch management module: backend consolidation layer, DynamoDB launchAsset table, Snowflake ingestion Lambda, and governance UI (TARC, DWG, GBT) with clinical trials timeline visualization
    Set up the team's AI coding workflow: Claude Code agents (task-splitter, pr-author, code-reviewer), Cursor skills, and Jira/Confluence automation via Atlassian MCP
    Stack: TypeScript, TanStack Start, React, Drizzle ORM, PostgreSQL, AWS Lambda, Pinecone, LaunchDarkly, SST v4, Claude Opus 4.6 via Bedrock
    Bedrock RAG Pinecone Claude Vercel AI SDK
  • Sanofi
    Sanofi · Software Engineer, Beyfortus Web Platform
    INDUSTRIE PHARMACEUTIQUE
    avril 2024 - avril 2025 (1 an)
    Paris, France
    Migrated the Beyfortus product website from a static site to a fully dynamic, CMS-driven platform.
    Integrated Magnolia CMS to enable editorial teams to manage content independently, eliminating deployment dependency for copy and page updates
    Rebuilt the frontend architecture in Next.js and TypeScript to support dynamic content rendering from the CMS API
    Stack: Next.js, TypeScript, Magnolia CMS
    Typescript NestJs magnolia Next.js MCP

Recommandations

Soyez le premier à recommander Jérôme

Contribuez à la réussite de ce freelance en partageant votre expérience de collaboration avec lui.

Ces profils de freelance correspondent également à vos critères

AgathaA

Agatha Frydrych

Backend Java Software Engineer

4.7

(3)

2

BaptisteB

Baptiste Duhen

Fullstack developer

4.6

(4)

5

AmedA

Amed Hamou

Senior Lead Developer

4

(2)

7

AudreyA

Audrey Champion

Web developer

4.3

(3)

4

Formations

  • Coding Academy By Epitech
    Epitech
    2016
    - Formation intensive de développeur web full Stack : - Connaissances en C et en C++. Permet ainsi une connaissance du fonctionnement de base du langage informatique. - Front-end : Javascript : Ember et React. HTML et CSS (Bootstrap) Back-end: Node JS.
  • Master de Communication
    Ircom
    2011

Compétences

Catégories