~/ Ren Okafor · Building intelligent systems
I design and deploy machine learning systems that solve real problems — from LLM fine-tuning and RAG pipelines to computer vision and production MLOps. I believe in models that work, not just models that impress.
A multi-agent LLM framework that generates, critiques, and iteratively refines long-form fiction. Uses a supervisor agent to coordinate specialized sub-agents for plot consistency, character voice, and tension pacing. Reduced hallucination rate by 62% over baseline GPT-4 via structured RAG + fine-tuning.
Real-time pedestrian and vehicle detection pipeline for smart city infrastructure. Runs at 48fps on edge hardware using custom YOLOv8 + TensorRT optimization.
Open-source MLflow alternative with built-in drift detection, cost tracking, and Slack alerting. Used by 2k+ engineers across 180 orgs.
Production-grade multilingual sentiment analysis API serving 2M+ requests/day. Fine-tuned XLM-RoBERTa on 18 languages with P99 latency under 80ms.
Published technique reducing LoRA memory footprint by 40% for resource-constrained environments. 380 GitHub stars, cited in 3 academic papers.
Available for full-time roles, contract projects, and research collaborations. I'm especially interested in applied LLM work and ML systems at scale.