Freelancer profile translated to English.
Back to original languageAbout Abderrahmane
AI Engineer | Agentic Systems, RAG & LLM Evaluation (Production-ready)
I design GenAI systems that actually go into production: business copilots, multi-step agents, and reliable RAG pipelines.
I intervene where most projects get stuck:
→ unreliable answers (hallucinations)
→ uncontrolled costs
→ difficult-to-operate systems
I work across the entire chain: architecture, optimization, and evaluation.
🔹 WHAT I DO
→ Transformation of GenAI POCs into production-ready systems
→ Business copilots (legal, ops, data) based on advanced RAG
→ AI Agents (decomposition, routing, multi-step reasoning)
→ Hybrid systems (vector search + SQL + business APIs)
→ Evaluation pipelines (reliability, accuracy, hallucinations)
→ Optimization under real constraints (latency, cost, scalability)
🔹 TECHNICAL EXPERTISE
/ Architecture
→ Advanced RAG (vector + BM25 + SQL routing)
→ Orchestrated agents (LangGraph, workflows)
→ Complex query decomposition
/ Optimization
→ API cost reduction (-40% via cache & routing)
→ Latency improvement (retrieval & prompts)
→ Retriever tuning (top-k, chunking, hybrid search)
/ Reliability & Evaluation
→ Metrics (RAGAS, contextual precision, recall)
→ Hallucination reduction
→ MLflow observability (tracking runs)
🔹 CONCRETE USE CASES
1.Legal Copilot (AMF-like)
→ Slow and unreliable search
→ Hybrid RAG + SQL + validation
→ Results:
Time ÷3
Traceable answers
Fewer business errors
2.Multi-step Research Agent
→ Complex multi-source questions
→ Decomposition + LangGraph
→ Results:
+25 to +40% accuracy
Better structured answers
3.RAG Pipeline Optimization
→ High costs + inconsistencies
→ Tuning + cache + reranking
→ Results:
-30 to -50% API cost
Improved relevance
French
Native or bilingual
English
Fluent
Arabic
Native or bilingual
Japanese
Basic
Can work on-site
Paris (up to 50km)
Experience
- Autorité desMarchés Financiers (AMF),GenAI Engineer (LLM, RAG & Agents)March 2025 - August 2025 (5 months)Paris, France• • Designed and deployed an LLM-based legal copilot for regulatory teams (document search, article analysis, response justification).• • Developed an agentic RAG pipeline (LangGraph, FAISS, reranking, cache) used by non-technical business users.• • +27% contextual accuracy, –60% latency, –18% factual errors through optimization of retrieval, embeddings, and caching.• • Implemented an evaluation and monitoring chain (MLflow, RAGAS) to track quality, costs, and model drift.• • Held workshops with legal experts and Data teams to prioritize use cases and secure business adoption.• • Packaged, documented, and transferred to operational conditions to the Data Intelligence team.
- Alstom,Automation & Reliability EngineerTECHSeptember 2023 - January 2024 (4 months)59300 Valenciennes, France• • Developed an industrial test automation tool (C#, Python, VBA) integrated into the MPU validation pipeline.• • –80% of test campaign time, improved traceability and auditability.• • Worked with systems and validation teams to ensure the reliability of a critical production deployment process.
Recommendations
Be the first to recommend Abderrahmane
Help this freelancer shine by sharing your experience working together.
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
- Embedded Systems EngineerINSA Hauts-de-France2026Ingénieur systèmes embarqués
- Academic ExchangeKeio University2025Échange académique