Principal AI Engineer | Architecting Governed & Local-First LLM Systems | SLM Researcher
I'm a Principal AI Engineer with 9+ years of experience bridging the gap between scalable software architecture and production-grade AI. I specialize in AI-Native Software Governance (Atomic Traceability & Preflight Enforcement) and Small Language Model (SLM) optimization for local-first, high-precision applications. Beyond engineering, I build SOTA translation baselines for underrepresented Nigerian languages and lead the Small Giants initiative to democratize efficient, private AI.
I am passionate about sharing knowledge through my YouTube channel and contributing to the open-source community, maintaining projects like OneCoder CLI and the Small Giants research initiative.
- 🤖 OneCoder CLI: Local-first sprint management and governance for coding agents and humans. Building tools for atomic traceability and preflight enforcement in AI-native workflows.
- 🤖 Small Giants: Building a curated collection of practical demonstrations showing how Small Language Models (SLMs) in the 350M-3B range can power production-grade AI applications locally.
- 🌍 TinyAya & Linguistic Diversity: Developing high-performance translation systems for underrepresented Nigerian languages (Yoruba, Hausa, Igbo, Pidgin) using
tiny-aya-earthand benchmarking with DSPy. - 🚀 Building Voice-First AI Systems: Exploring human-computer interaction with AI Research CoPilot—a voice-first assistant using Pipecat and Gemini.
- AI/ML: Python, PyTorch, LangChain, LangGraph, DSPy v3+, RAG, Qdrant, SLMs (Liquid AI, IBM Granite, TinyAya)
- Backend: Node.js, Java (Spring Boot), RESTful APIs, Microservices, MCP
- Frontend: Next.js, Angular, TypeScript, Streamlit
- Cloud & DevOps: AWS, GCP, Azure, Docker, Kubernetes (CKAD), Terraform, uv (Python Package Manager)
- Databases: PostgreSQL, MongoDB, Redis, MySQL
- OneCoder CLI: A governance-first CLI for AI-assisted development.
- Atomic Traceability: Links commits to Sprint-IDs and Spec-IDs for clear provenance.
- Preflight Enforcement: Validates task breakdowns, LOC limits, and documentation before progress.
- Local-First Architecture: Metadata stored within
.sprint/and.issues/for repository-centric management.
- Small Giants: A research-led initiative for local-first AI.
- Invoice Parser: Multimodal extraction using Liquid AI LFM2-VL.
- Granite Coder: Token-efficient coding agent with recursive language models.
- Local RAG Pipeline: semantic retrieval using LangChain, Qdrant, and Ollama, with automated evaluation via RAGAS.
- TinyAya Translation MVP: A translation system for EN ↔ YO/HA/IG/PCM with CLI and Gradio UI, rigorously evaluated using DSPy metrics.
- AI Research CoPilot: A voice-first AI assistant built with Next.js, Pipecat, and Gemini 2.5 for real-time research retrieval.
- SpeechWhisper (macOS App): A native Swift application for real-time, on-device speech-to-text transcription.
- Winner - GenHax Hackathon (2024)
- Participant & Builder - AI Tinkerers Gemini x Pipecat Virtual Hackathon (2025)
- Certified: Azure AI Fundamentals, Google Cloud Developer, Kubernetes Application Developer (CKAD)