Brell SANWOUO
Brell SANWOUO PhD
MSc. Artificial Intelligence

PhD Candidate @ Inria Lille / University of Lille

Self-Adaptive System · Autonomous System · AI · Multi-agent System Modeling · World Model · Digital Twins

Recent Highlights

Research Focus

I am a PhD student at Inria Lille and the University of Lille, where I work at the intersection of AI and Software Engineering (AI4SE and SE4AI). My goal is to design methods that make software systems and software engineers more resilient, adaptive, and efficient.

My research covers software architecture, self-adaptive and autonomous systems, AI agents, and empirical software engineering. A core part of my work is the modeling of multi-agent systems, software system structures, and behavioral models for adaptation and autonomy at runtime.

I am also interested in reliable and reproducible AI for Software Engineering, with a strong focus on innovation and on translating research results into concrete, real-world software engineering practices.

Keywords AI4SE SE4AI Self-adaptive systems Autonomous systems AI agents Multi-agent systems Runtime adaptation

Selected Publications

Dynamic Agent Generation for Self-Adaptive Root Cause Analysis

Brell SANWOUO, Clément Quinton, and Paul Temple, SEAMS 2026 (Rank A) . April 2026.

PDF BibTeX
Artifact of Dynamic Agent Generation for Self-Adaptive Root Cause Analysis Best Artifact Award

Brell SANWOUO, Paul Temple, and Clément Quinton, SEAMS 2026 (Rank A) - Artifact Track . April 2026.

Artifact BibTeX
Generative AI-based Adaptation in Microservices Architectures: A Systematic Mapping Study

Brell SANWOUO, Clément Quinton, and Paul Temple, ICWS 2025 (Rank A) . July 2025.

PDF BibTeX
View all publications

Projects

AWARE

Research Artifact

A minimal Assess prototype for telemetry-based root-cause analysis. It combines a Parser agent, an Executor agent, and dynamic analysis sub-agents to turn an incident question into a validated BuildSpec and an execution report.

Root-cause analysis Telemetry Multi-agent systems BuildSpec
Open project View all projects

AgentLantern

Developer Tool

A CLI devtool that documents, lints, inspects, plays, and replays multi-agent AI systems across major Python and JavaScript agent frameworks.

Multi-agent AI CLI Documentation Live replay
Open documentation View source

GEAR Framework

Research Framework

A framework-independent design layer for portable multi-agent systems, using feature-model-based specifications to generate configurations and executable code for multiple agent frameworks.

Portable MAS Feature models Code generation Variability
View source

Agent Energy Consumption

In Progress

An ongoing research project studying how to observe and measure the energy consumption of multi-agent AI systems across different execution configurations.

Sustainable AI Energy measurement Multi-agent systems Observability

Open Knowledge · Learnix

Open Learning Platform

A free learning platform offering structured paths, open courses, and practical notebooks for Python, data science, and machine learning.

Open knowledge Python Data science Machine learning
Open platform

Talks & Presentations

2026

SEAMS — Research Track · Dynamic Agent Generation for Self-Adaptive Root Cause Analysis

Presented in Rio de Janeiro, Brazil.

SEAMS — Artifact Track · Artifact of Dynamic Agent Generation for Self-Adaptive Root Cause Analysis Best Artifact Award

Presented in Rio de Janeiro, Brazil.

2025

ICWS · Generative AI-based Adaptation in Microservices Architectures

Presented in Helsinki, Finland.

FSE · Breaking the Loop: AWARE is the New MAPE-K

Presented in Trondheim, Norway.

2024

BENEVOL · Toward AI-based Complex Self-Adaptive Systems

Workshop poster presented in Namur, Belgium.

Awards & Honours

SEAMS 2026 - Best Artifact Award

Best Artifact Award

Artifact Track · Tue 14 Apr 2026 · Rio de Janeiro, Brazil

Received the Best Artifact Award for Artifact of Dynamic Agent Generation for Self-Adaptive Root Cause Analysis at the SEAMS 2026 Artifact Track.