Abidjan, Côte d’Ivoire · LinkedIn · Email
I design governed data, knowledge and decision-support systems that transform heterogeneous datasets and documents into traceable, searchable and actionable knowledge.
Je conçois des plateformes de données, de connaissances et d’aide à la décision qui transforment des sources hétérogènes en informations structurées, traçables, recherchables et exploitables.
Judicial administrator, project manager and Data & AI Governance practitioner with 12+ years of experience in Côte d’Ivoire’s justice sector. Current work combines public-sector transformation, data engineering, document intelligence, information retrieval, local AI and scientific evaluation.
Administrateur des services judiciaires, chef de projet et praticien de la gouvernance des données et de l’IA, avec plus de douze années d’expérience dans le secteur de la justice en Côte d’Ivoire. Mes travaux relient transformation publique, data engineering, intelligence documentaire, recherche d’information, IA locale et évaluation scientifique.
Current positions:
- Deputy Director of Studies and Statistics, Ministry of Justice, Côte d’Ivoire
- Head of IT and Statistics Services, Court of Cassation of Côte d’Ivoire
- Justice Coordinator, USAID ProJustice-CI Program
- Project and Program Manager for data, AI and digital-transformation initiatives
- PhD-track researcher in Judicial Data Governance, Digital Justice and Responsible AI
- Le Wagon Data Science & AI graduate, Batch 2149 (2025)
Technology alone does not transform institutions. Governed data, institutional learning and responsible innovation do.
Independent, domain-neutral and local-first document intelligence platform. LIAOS combines governed ingestion, SQLite storage, document provenance, citations, lexical retrieval, reliable workflows and local language models.
Key results:
- 2,276 indexed documents and 143,746 traceable chunks
- TXT, Markdown, HTML, PDF and DOCX ingestion
- SimpleSearch and SQLite FTS5/BM25 retrieval engines
- transactional outbox, idempotency, dead-letter workflows and persistent scheduling
- governed source registry and secure bounded collection
- Ollama and LM Studio integration without mandatory cloud dependency
- reproducible information-retrieval evaluation with Goldsets, Hit@k, Precision@k, Recall@k, MRR and nDCG
- approximately 396 automated tests
Current milestone: human annotation of the scientific Goldset before the official SimpleSearch and FTS5 baselines.
Modular governance and control framework for AI-assisted projects, knowledge systems and supervised workflows.
- versioned manifests and JSON Schema validation
jdgfcommand-line interface- Knowledge, Advanced RAG, Agent and Workflow contracts
- security-by-default and human-approval gates
- optional PostgreSQL, Redis, Neo4j, Qdrant, Ollama and LiteLLM contracts
- FastAPI control surface, CI hardening and dependency auditing
- current developer preview: v0.10.1
Doctoral funding intelligence and decision-support platform.
- opportunity, evidence and application-readiness models
- deterministic scoring and supervisor matching
- Streamlit cockpit and FastAPI-compatible interfaces
- SQLite persistence, Snowflake preparation and cloud deployment workflows
- mandatory human validation before outreach or application submission
Digital-learning and knowledge-governance prototype for court registries and judicial institutions.
- structured judicial knowledge corpus and source inventories
- Streamlit learning prototype
- governance-state, ownership and validation models
- documented institutional pilot use cases
Related project: CS GREFFE OS V2
Reproducible Python data-engineering pipeline for heterogeneous judicial HR, infrastructure and operational Excel workbooks.
- extraction, normalization and row-level provenance
- CSV and Excel delivery contracts
- data-quality diagnostics and validation
- CLI migration from exploratory notebooks
- automated tests, visualizations and Power BI preparation
AI-assisted local commercial concierge prototype that turns an unstructured user need into explainable product or service recommendations.
- Streamlit and SQLite MVP
- needs extraction and vendor scoring
- price, quality, availability and trust criteria
- human verification, administrative cockpit and Docker preparation
Private knowledge base and long-term documentary memory for research, projects and validated decisions, designed for read-only ingestion by governed knowledge systems.
Main theme: Data Governance and Digital Transformation of Judicial Systems in Africa
Research question: How can data-governance frameworks support the digital transformation and AI readiness of judicial systems in African public institutions?
Focus areas:
- Judicial Data Governance
- Digital Justice and judicial information systems
- Responsible AI and AI Governance
- Institutional intelligence and knowledge management
- Information retrieval and document intelligence
- AI readiness of African courts
- Evidence-based justice policy
- Human-centered public-sector transformation
Python · SQL · pandas · SQLite · ETL · Excel and CSV processing · data quality · metadata · lineage · provenance · document ingestion · Unicode normalization · chunking · deduplication · Power BI · Snowflake
Full-text search · SQLite FTS5 · BM25 · document citations · Goldsets · human annotation · Hit@k · Precision@k · Recall@k · MRR · nDCG · reproducible baselines · SHA-256 corpus controls
FastAPI · Streamlit · REST APIs · persistent schedulers · leases · retries · exponential backoff · transactional outbox · idempotency · dead-letter queues · audit logging · recovery workflows
Ollama · LM Studio · local language models · prompt engineering · RAG foundations · AI Governance · Data Governance · human-in-the-loop · source authorization · robots.txt · least privilege · explainability
Git · GitHub Actions · Docker · Docker Compose · Linux · CI/CD · VPS deployment · Streamlit Cloud · automated testing · pytest · dependency auditing
DAMA-DMBOK · PMBOK · COBIT · TOGAF · ISO 8000 · ISO 27001 · Knowledge Management · Change Management · Human-Centered Design
Institutional framework for data quality, metadata, accountability, maturity assessment and AI readiness in African judicial institutions.
Data pipelines and methodologies supporting national judicial statistics and evidence-based justice policy.
Industrial anomaly-detection machine-learning pipeline developed during Le Wagon Data Science & AI, with FastAPI, Streamlit, Docker and GCP.
Research on Judicial Data Governance, Responsible AI and AI-ready public institutions in Africa.
- Official Judicial Statistics Repository
- Judicial and Penitentiary Statistics Yearbook 2021–2022
- Judicial and Penitentiary Statistics Yearbook 2019–2020
- Judicial and Penitentiary Statistics Yearbook 2017–2018
Working paper: Data Governance as Foundational Infrastructure for Intelligent Justice Systems in Africa
- PhD supervision and co-supervision in Judicial Data Governance and Responsible AI
- Data Engineer, Knowledge Engineer, Python Developer and Research Engineer opportunities
- Digital Justice and public-sector data-governance programs
- Information-retrieval, document-intelligence and local-AI projects
- Applied research on AI readiness, human oversight and institutional knowledge
- International cooperation on rule of law, digital transformation and evidence-based reform
Data Engineer · Knowledge Engineer · Python Developer · Research Engineer · Backend Engineer · Data Governance · AI Governance · Information Retrieval · Document Intelligence · ETL · Data Pipeline · Data Quality · SQL · SQLite · FTS5 · BM25 · FastAPI · Streamlit · pandas · Power BI · Snowflake · Docker · GitHub Actions · CI/CD · Linux · Local LLM · Ollama · LM Studio · RAG · Automated Testing · Software Architecture · Human-in-the-loop · Responsible AI · Digital Justice