Required skills
- Retrieval-Augmented Generation (RAG)
- LangChain
- LlamaIndex
- Vector Databases
- Large Language Models (LLM)
- Generative Artificial Intelligence
- Semantic Search
- Hybrid Search
- Reranking
- Python
- Data Processing
- Document Ingestion Pipelines
Languages: French
Job description
Seeking an AI Agent Governance Expert to build the knowledge base and RAG engine for the Architecture Copilot use case, focusing on ingesting and indexing technical documentation while ensuring accurate, sourced responses.
Project overview
Context
Intervention within the AI Prod Directorate, as part of the agentic foundation, for the benefit of our client's Architecture Department.
Missions
Working Conditions
Expected Deliverables
Technologies / Skills & Soft Skills
Technical Environment
Soft Skills
- Construction of the DRC-DA AI agent knowledge base and the RAG engine for the first priority use case, the Architecture Copilot
- Identification, ingestion, and indexing of DRC technical documentation
- Guarantee of sourced, verifiable responses compliant with document access rights
- Measurement of response quality with pilot architects
- Identification of documentary sources with DRC-DA referents: network standards and repositories, DAT, DAC, HLD, LLD, reference architectures, procedures, configurations, reports, and feedback
- Evaluation of the quality, volume, formats, freshness, and access rights of each source
- Feasibility study of the RAG approach on DRC documentation
- Identification of technical prerequisites and IT dependencies necessary for source access
- Implementation of ingestion, chunking, metadata enrichment, and document indexing pipelines
- Processing of Word and PDF documents, diagrams, tables, flow matrices, addressing plans, and configurations
- Design of the search strategy: semantic search, hybrid search, reranking, and scope filtering
- Management of access rights, lifecycles, and updating of documentary content
- Guarantee of responses accompanied by precise references to the documents and passages used
- Detection and reporting of missing or contradictory information
- Implementation of mechanisms to prevent the invention of configurations, references, or technical decisions
- Construction of reference question sets with pilot architects
- Measurement of response relevance, accuracy, and citation rate
- Contribution to the generation of initial architecture deliverables: first version of DAT, prerequisites, and risk analyses
- Documentation of technical choices, procedures, and implemented practices
- Contribution to skills transfer to network architects, AI teams, and security teams
- Integration of work into an industrializable trajectory
- Collaboration with DRC-DA architects and referents, the AI Prod Directorate, the AI Product Owner, the security referent, and pilot architects
- 3 months renewable
- Location: 93210 Saint-Denis
- Max purchase TJM: €600
- DRC documentary sources mapping
- RAG approach feasibility study
- Indexed DRC knowledge base
- Documented ingestion and update pipelines
- Chunking, metadata enrichment, and indexing strategy
- Semantic, hybrid search, reranking, and scope filtering strategy
- Architecture Copilot RAG engine
- Source citation mechanisms
- Document access rights compliance mechanisms
- Reference question sets
- Response quality evaluation report
- Measurement of relevance, accuracy, and citation rate
- Technical documentation of the knowledge base
- Operating and update procedures
- First version of architecture deliverables generated by the Copilot
- Documentation of prerequisites and risk analyses
- Updated tracking table
- Steering committee support materials and reports
- Skills and knowledge transfer materials
- RAG (Retrieval-Augmented Generation)
- LLM Frameworks: LangChain, LlamaIndex, or equivalent
- Vector Databases
- Generative Artificial Intelligence
- LLM
- Semantic Search
- Hybrid Search
- Reranking
- Python
- Data Processing
- Document Ingestion Pipelines
- PDF and Word Documents
- Diagrams, tables, and flow matrices
- Addressing plans and network configurations
- Documentary metadata
- Data Governance
- Access Rights Management
- Data Security
- Kubernetes
- Cloud Environments
- Network and Telecom Environments
- Critical Information System Architectures
- DRC Documentation: DAT, DAC, HLD, and LLD
- Rigor and service orientation
- Excellent analytical and synthesis skills
- Autonomy and ability to take charge of complex subjects
- Aptitude for production AI topics and agentic engineering
- Proactive and taste for continuous improvement
- Interpersonal skills with network architects, AI teams, and security teams
- Ability to collaborate with multiple technical and business stakeholders
- Pedagogical skills and ability to transfer practices
- Strong sensitivity to data confidentiality, security, and governance issues
- Ability to evolve in a complex and innovative environment
- Writing skills and ability to produce structured documentation
Project details
- Start: flexible
- Remote: hybrid
- Location: Saint-Denis, France, FR
- Rate: 75-75 EUR/h
- Languages: French
Indexed by 9am's meta-search. You are a job provider? Claim or manage this listing
Found on collective.work