Senior AI Engineer (RAG/Search, LangChain)

We are hiring a Senior AI Engineer to build the core AI system behind a high-impact project: digitizing law in Morocco and Africa and creating the first AI-queryable legal knowledge base.

Our goal is to deliver reliable, source-grounded answers to legal questions, based on a large corpus of heterogeneous legal documents.

🚀 Why this project

• Build a legal knowledge infrastructure, not a demo
• Work on high-precision AI where accuracy, traceability, and reliability matter
• Create a long-term asset (Morocco → Africa) with real societal impact

🎯 Your mission

Own the end-to-end AI pipeline:

Legal documents → structured data → retrieval → cited answers

You will design, build, and improve both the document processing pipelines and the RAG/search system, with a strong focus on quality, evaluation, and robustness.

🔧 Key responsibilities

• Process and structure heterogeneous legal documents (PDFs, scans, Word, images):
  • parsing, OCR, cleaning, structuring (articles, sections, hierarchy)
  • intelligent chunking with full traceability and metadata
• Build and improve retrieval & RAG:
  • hybrid search (BM25 + embeddings)
  • reranking (cross-encoders)
  • query understanding and expansion
• Ensure grounded, reliable answers:
  • strict citation to sources
  • no-answer / controlled refusal when evidence is missing
• Measure and improve quality:
  • evaluation datasets, relevance metrics (Recall@k, MRR, nDCG)
  • regression testing and error analysis
• Production readiness:
  • robust pipelines, monitoring, logging, performance

✅ Profile

• 5+ years of experience in AI / NLP / Search / Data Engineering roles
• Strong Python skills
• Hands-on experience with unstructured documents and/or RAG systems
• Strong culture of measurement and experimentation
• Professional proficiency in English, French is a plus.

🛠 Tech stack (indicative)

• Document processing & OCR tools
• Retrieval: OpenSearch / Elasticsearch, vector search (FAISS, pgvector, etc.)
• Reranking: HuggingFace / cross-encoders
• RAG orchestration: LangChain or LlamaIndex (nice-to-have)
• Cloud: AWS