Data Engineer — I design and deploy data pipelines and AI systems, from proof of concept to production.
Currently completing a Data Engineering master's program at OpenClassrooms, I'm building end-to-end experience across the full data lifecycle: NoSQL modeling, ELT pipelines, real-time processing, workflow orchestration, and RAG/NLP systems deployed on the cloud. Each project below stems from a concrete professional brief and demonstrates a complete, hands-on technical skill set.
Languages & processing · Python · PySpark · SQL · Polars Databases · PostgreSQL · MongoDB · pgvector · Redis · FAISS Pipelines & orchestration · dbt · Airbyte · Kestra · Redpanda (Kafka) Cloud & deployment · AWS (RDS, ECS Fargate, ECR, S3, EventBridge, Bedrock, ElastiCache, CloudWatch) · Docker · GitHub Actions AI / NLP · LangChain · Mistral AI · RAG · smolagents · RAGAS
Each project addresses a realistic professional brief and is documented with a detailed README (context, architecture, stack, key takeaways).
Design study to industrialize a RAG chatbot (Puls-Events): scalable cloud architecture, prioritized macro backlog, build/OPEX cost modeling. Stateless architecture on AWS — all state held in managed services (RDS + pgvector, ElastiCache, Bedrock, ECS Fargate), with agentic web search via smolagents.
Architecture · AWS · Project management · Cost modeling
Proof of concept for a sports-activity rewards system at Sport Data Solution (fitness analytics start-up): a full pipeline covering extraction, transformation, loading, data-quality testing, and continuous monitoring. Computes the financial impact of employee sports incentives, with Slack notifications and a Power BI report featuring replayable historical indicators. ELT · Data quality · Monitoring · Power BI · Slack 🔗 [P12 repo link]
Proof of concept for a retrieval-augmented recommendation engine: LangChain + Mistral + FAISS, fed by the OpenAgenda API and evaluated with RAGAS. Surfaced the limits of RAG (metadata filtering, the living dataset problem).
RAG · LangChain · Mistral · FAISS · RAGAS
🔗 [P11 repo link]
Design of an orchestrated pipeline architecture integrating Spark, Redpanda, and PostgreSQL. Study of declarative orchestration and dependency management between tasks.
Kestra · Orchestration · Spark · PostgreSQL
🔗 [P10 repo link]
Streaming proof of concept with PySpark Structured Streaming and Redpanda in a local Docker environment. Modeling of a hybrid infrastructure for an industrial data use case.
PySpark · Structured Streaming · Redpanda · Docker
🔗 [P9 repo link]
End-to-end ELT pipeline: Airbyte (ingestion) + PostgreSQL + dbt Core (transformation), deployed on AWS (RDS, ECR, ECS Fargate, EventBridge). Modern Extract-Load-Transform approach with dbt tests and documentation.
ELT · dbt · Airbyte · AWS · PostgreSQL
🔗 [P8 repo link]
Analysis of rental data (Airbnb Paris/Lyon) with MongoDB: ReplicaSet, sharding, aggregation pipelines, Python integration via PyMongo and Polars.
MongoDB · NoSQL · Aggregation · Sharding
🔗 [P7 repo link]