I am an engineer in Artificial Intelligence and Data Science, with a GPA of 17/20. I am interested in both AI research and software development. I have worked with LLMs, AI agents and transformer models, including projects using knowledge graphs and different types of technical data. I am also working on Memsash, a personal research project where I explore new ways to give AI agents long-term memory. My main interests are AI agents, knowledge representation, reasoning and explainable AI.
Experiences
Software, DevOps & AI Engineer
Crédit Agricole – Montrouge, France
Dec 2024 → Present
Design and develop enterprise software and BPMN-based workflow systems for leasing and factoring platforms within Crédit Agricole. Contributed to the design and industrialization of an AI agent for BPM process analysis, incident investigation, and developer onboarding, using LLMs, LangGraph, MCP, Qdrant, and Neo4j to connect process information with technical documentation and business specifications. Build and operate scalable services and CI/CD pipelines using Java/Spring Boot, Python, Kubernetes, and GitLab CI in a regulated production environment.
View details: Software, DevOps & AI EngineerSoftware and AI/Data Engineer
ADEO – Lille, France
Sep 2023 → Sep 2024
Designed and developed data and AI systems for analyzing customer activity across ADEO business units. Built an LLM-powered Text2SQL agent enabling natural-language interaction with enterprise databases, from user queries to executable SQL. Developed scalable data pipelines and contributed to a Spring Boot CIAM application and data visualization solutions.
View details: Software and AI/Data EngineerSoftware Engineering Intern
ADEO – Lille, France
Jun 2023 → Sep 2023
Developed a multi-tenant administration console for OAuth2 Client IDs management across Reachfive instances. Implemented frontend and backend components with Next.js, RBAC, and automated deployment pipelines.
View details: Software Engineering InternResearch
MemHash: Associative Hash-Addressed Memory for AI Agents
Investigating a novel memory architecture for AI agents combining semantic hashing, contextual associations, and sparse memory addressing. The research explores direct or near-direct access to associated memories beyond vector similarity search, including memories linked through experience or explicit relationships. Current research directions include sparse routing, knowledge-graph-augmented associations for structured knowledge representation, and scalable control of memory collision
Paper · In Progress
- AI Agent Memory
- Locality-Sensitive Hashing (LSH)
- Knowledge Graphs - GraphRAG
- Sparse Distributed Memory
Do Large Language Models Change How Developers Seek Help? A Trace-Based Study of Stack Overflow Behavior
Led and initiated a research project in collaboration with Anna Baj-Rogowska investigating the impact of Large Language Models (LLMs) on observable problem-solving behaviors in technical online communities. Defined the scientific research question, designed the methodological framework, formulated the research hypotheses, and structured the large-scale statistical analysis using Stack Overflow data (2019–2025) to study changes in self-regulated learning behaviors before and after the rise of LLMs. The paper was accepted at ISD 2026, where I presented our work in Prague in September.
Paper · Published · ISD 2026
- LLM-Assisted Data Annotation
- Self-Regulated Learning
- Mistral AI - Batch API
- SQL
Ethics of Conversational AI
Conducted a literature review on the ethics of conversational AI, focusing on technical aspects, regulatory frameworks, and best practices. Analyzed biases, transparency, and accountability in AI models like ChatGPT, with a structured approach praised by the defense jury.
Literature Review
- AI Ethics
- AI Bias
- Responsible AI
- AI Governance
Do Social Media Virtual Communities Support Purchase Decision Making?
In collaboration with Anna Baj-Rogowska (Assistant Professor, Poland), we investigated the influence of virtual communities on the purchasing decision-making process, with a particular focus on the analysis of comments posted under YouTube influencer videos. The paper was accepted at the ISD 2025 International Conference, where I had the opportunity to present our findings in Serbia.
Paper · Published · ISD 2025
- Topic modeling
- BERTopic
- Dimensionality Reduction (UMAP, PCA)
- Clustering
LLMFSD: LLM Fake Structured Data, faking structured data from any LLM
Proposed an innovative approach to generate fake structured data using LLMs. The project enables developers to create synthetic datasets via SQL-like queries, bridging the gap between unstructured LLM outputs and structured data requirements for testing and prototyping. Developed a Python package (llmfsd) available on PyPI.
Software
- Large Language Models
- Structured Data Generation
- Prompt Engineering
- SQL
Projects
Cvolio: AI-Powered CV & Portfolio Platform
Jan 2026
Cvolio is a CV and portfolio builder that uses AI to understand existing CVs, turn them into structured data, improve the content, and adapt it to a job offer. Building it gave me practical experience with LLM integration, structured outputs, prompt design, and adding AI features to a real web application.
Ubenia: Conversational AI for Beninese Higher Education
Oct 2025
Leading the development of an ongoing conversational AI platform that enables users to query information about universities and higher education in Benin through a ChatGPT-like interface. The system uses MCP and LangGraph to orchestrate AI agents, Qdrant for knowledge retrieval, Hugging Face models and Mistral for LLM-based processing, and Neo4j for knowledge representation. A network of university ambassadors contributes to collecting and validating institutional data used to continuously enrich the system’s knowledge base.
In Progress
xrMCP: Define MCP tools once, deploy everywhere
Apr 2026
Building an open-source project that makes MCP tools easier to create, install and share. Tools are defined as portable manifests and can be installed from a public registry instead of building a separate MCP server for every integration. The project includes Go and Python SDKs, a CLI, tool packaging and a public registry, with releases available for Windows, Linux and macOS.
Transformers from scratch (Attention Is All You Need)
May 2024
Built the original Transformer architecture from scratch, with multi-head attention, positional encoding, encoder-decoder blocks, masking, residual connections, and feed-forward networks. This project helped me better understand how Transformers work and how they are used in modern language models.
Fine-tuning GPT-2 for playing chess
Aug 2024
QR Code generator
Mar 2023
Data pipeline (Kafka, PubSub, dbt, GCP, Terraform, ...)
Feb 2024
Orders analysis (Airflow, dbt, Kafka, MinIO, Spark, ...)
Aug 2024
dddb: NoSQL Lite DB (C)
Feb 2024
Education
École d’Ingénieurs du Littoral - Côte d’Opale (EILCO), France
Sep 2021 – Sep 2024
Engineering degree (BAC+5) in Data Science, Data Engineering and Software Development (specialization in AI, GPA: 17/20)
Haute École Libre Mosane (HELMo), Belgium
Feb 2023 – Jun 2023
Cybersecurity - Erasmus mobility
Institut de Mathématiques et de Sciences Physiques (IMSP), Benin
Sep 2020 – Sep 2021
Bachelor's degree in Computer Science (Very good, valedictorian)
Institut de Mathématiques et de Sciences Physiques (IMSP), Benin
Sep 2018 – Sep 2020
Scientific Preparatory Classes (Very good, valedictorian)