I bully models out of notebooks and into production—agents, inference, evals, computer vision, and whatever breaks after deploy.
- AI Engineer at Octave Inc — developing LLM agents, FastAPI services, and production AI systems.
- AI Engineer at Gesture AI — building computer-vision systems and optimizing models for edge deployment.
- Technical Lead at AISOC — leading workshops, community initiatives, and knowledge transfer.
- Computer Science student at the University of Ibadan — exploring reliable agents, inference systems, evaluation, AI safety, and MLOps.
I like opening the black box, understanding why it works, and rebuilding the important parts from first principles.
LANGUAGES & CORE
ML & DEEP LEARNING
AGENTS & LLM SYSTEMS
BACKEND, DATA & DEPLOYMENT
A few systems I built to understand—and improve—the machinery behind production AI.
|
Implements KV caching, static and continuous batching, speculative decoding, and a FastAPI serving layer for Qwen2.5-0.5B. ~3× faster generation with KV caching
|
Runs untrusted AI-generated code inside an isolated runtime with cgroups v2, network air-gapping, resource limits, and watchdog timeouts. 7 layers of runtime isolation
|
|
Combines golden datasets, deterministic checks, LLM-as-judge scoring, reports, and regression tracking in one reproducible harness. 5 judge failure modes reproduced and documented
|
A voice-first shared-object memory system with natural corrections, guided Lost Mode, shared custody, and privacy-aware access. 6-tool Streamable HTTP MCP interface
|
Public GitHub activity · private contributions are excluded.
B.Sc. Computer Science
University of Ibadan · Expected 2027
- Introduction to Deep Learning — MIT 6.S191 · Completed
- Supervised Machine Learning — Coursera · Completed
Always building. Always learning.