Freelancer profile translated to English.
Back to original languageAbout Bakary
Senior LLMOps & MLOps Engineer (5+ years) – Specializing in RAG systems and generative AI in production (AWS, GCP)
I design and deploy AI systems in production (RAG, LLMs, ML) for critical use cases, with a focus on performance, scalability, and cost optimization.
Core Expertise:
• Production-ready RAG architecture (hybrid search, reranking, vector DB: Qdrant, Pinecone)
• LLM Industrialization (vLLM, GPU deployment, scalable APIs)
• End-to-end MLOps (CI/CD, Prefect/Airflow orchestration, monitoring & drift)
Impact:
• Improved accuracy of semantic search systems
• Reduced LLM inference latency and costs
• Deployment of robust and automated ML pipelines
Stack:
LLMOps: LangChain, LangGraph, vLLM, Hugging Face
Vector DB: Qdrant, Pinecone, FAISS
Cloud: AWS (SageMaker, S3, Lambda), GCP (Vertex AI)
Data/ML: PyTorch, XGBoost, PySpark
Available for assignments:
• RAG / Generative AI
• MLOps / ML Industrialization
• Optimization of LLM systems in production
French
Native or bilingual
English
Fluent
Can work on-site
Choisy-le-Roi (up to 50km)
Experience
- SanofiAI & Data EngineerHEALTH AND WELLNESSMarch 2026 - Today (5 months)Paris 16 Passy, FranceMember of the SILC team (Sanofi Integrated Launch Capabilities), a platform supporting Sanofi brand teams in launching pharmaceutical products. Contribution to SILC Agent, a GenAI assistant: a multi-tool RAG agent on AWS Bedrock (Claude Opus) and Pinecone, fed from SharePoint, assisting teams with Situation Analysis, Target Product Profile, competitive intelligence, KPI tracking, and PowerPoint generation. GxP pharma environment, multi-environment CI/CD.🔹 Vector Lifecycle ManagementDesign and delivery of Pinecone vector lifecycle management (create/update/rename/delete) for the agent's knowledge base, as well as the ingestion pipeline.🔹 LLM Evaluation Framework (offline + online)Design and delivery of the SILC Agent's quality evaluation system with W&B Weave (LLM-as-judge).🔹 Bedrock AgentCore + Sanofi Agent Hub + A2A MigrationContribution to migrating the SILC Agent from TypeScript/Vercel AI SDK (silc-frontend) to a containerized Python runtime on AgentCore, multi-agent orchestration (Strands), tool calling, and registration on the Sanofi Agent Hub (AGENTCORE + A2A modes).🔹 Team OperationsScrum (SILC GenAI Squad, 2-week sprints), design reviews, pair code reviews, incident response, knowledge transfer.Stack: Python, AWS Bedrock (Claude Opus/Sonnet), Pinecone, W&B Weave, Bedrock AgentCore, Strands, Multi-agent, Tool calling, A2A, RAG, LLM-as-judge, AWS (Lambda, Step Functions, SQS, SNS, DynamoDB, EventBridge), Terraform, GitHub Actions, Snowflake, dbt, Streamlit
- Base Claude BernardLead Data & AIHEALTH AND WELLNESSSeptember 2025 - February 2026 (5 months)Île-de-France, France• Design and deployment of a medical RAG system in production (880k+ documents), improving response accuracy and ensuring traceability (sourced citations)• Implementation of a multi-step retrieval pipeline:- Query validation via LLM (medical filtering)- Multi-query expansion (semantic coverage)- Hybrid search (Qdrant: dense BGE-M3 + sparse Splade + fusion)- Reranking via cross-encoder for high clinical accuracy• LLM Industrialization:- vLLM deployment (OpenAI-like API) on GPU (RunAI)- Asynchronous processing (Celery/Redis) → latency reduction• Implementation of a production LLMOps stack:- GitLab CI/CD, Docker containerization, monitoring- Reliability, reproducibility, and cost optimization• Development of incremental data pipelines (Prefect):- Multi-source ingestion (ANSM, HAS…)- Intelligent versioning (hash) → reduction of recalculationsStack: Qdrant, vLLM, LangChain, Prefect, FastAPI, Docker, GitLab CI/CD, PostgreSQL
- LisiData & MLOps EngineerCONSULTING AND AUDITSNovember 2022 - Today (3 years and 8 months)Paris, France
- • Development of end-to-end MLOps pipelines on AWS SageMaker:
- GitLab CI/CD, automated deployment, model registry- Drift monitoring with alerts → robustness improvement- Design of scalable data pipelines:
- ETL (AWS Glue, PySpark), orchestration (Airflow)- S3 Data Lake + analytics (Athena)- Development of APIs and access security:
- AWS Lambda + API Gateway- User management via Cognito- Implementation of an industrial RAG system:
- Semantic search on technical documentation- Observability (LangSmith): latency, costs, hallucinationsStack: AWS (SageMaker, Glue, Airflow, Lambda, S3), LangChain, OpenSearch, PySpark
Recommendations
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- Master 2, IA and Data ScienceÉcole Polytechnique2020Master 2, Data Science
- Master 2 (M2), Multimedia NetworkingTélécom ParisTech2019Master 2 (M2), Multimedia Networking
Certifications
- AWS Machine Learning SpécialityAmazon Web Services (AWS)2023