About me
A Burundian ML & AI Engineer based in South Korea, studying Smart Computing at Kyungdong University (KDU Global) — working at the intersection of applied AI, systems thinking, and civic innovation.
My work spans from deep learning experiments to full-stack platforms designed to serve communities. On one side, I focus on building real-world products that leverage existing trained models through APIs — turning AI capabilities into tangible, accessible tools. On the other, I pursue a deeper understanding of what happens beneath the surface — how models learn, how they reason, and how their outputs can be interpreted and trusted.
This dual focus shapes everything I build: systems that are not just functional, but grounded in how the technology actually works.
Recent work includes ArrivAI, a migrant economic integration platform for South Korea, and MGM Assist, a civic intelligence platform built during a GenAI Academy hackathon. Details and demos live on the Projects page.
I also compete in AI challenges that stress-test reasoning and delivery under constraints. Voice Escalation via Vocal Bridge — ranked #3 of 38 builders in the DeepLearning.AI × Vocal Bridge 7-Day Voice AI Builder Challenge (June 2026) — and ARC-AGI-2, an ongoing Kaggle benchmark on abstraction and reasoning from minimal examples. Write-ups and demos are on the Challenges page.
Beyond shipping, I keep returning to a few questions:
- How AI can serve underrepresented communities without reducing them to edge cases
- The gap between model fundamentals and field deployment
- Products people actually revisit
- Learning quickly from both wins and dead ends
I am actively collaborating with founders, civic leaders, and researchers who care about measurable impact.
Outside of code: reading on compound systems, exploring South Korea, and sketching the next problem worth a deep build.
If you are working on civic AI, migrant economic integration, or trustworthy decision systems, I would like to hear from you.