I am a Software Engineer and AI/ML Engineer operating at the intersection of high-concurrency Go backends and Python AI/ML infrastructure. My work focuses on building low-latency distributed microservices alongside local LLM routers, multi-agent systems, and medical computer vision pipelines.
Having completed the MIT Open Learning "Universal AI" curriculum under 3MTT, I combine low-level systems engineering with modern AI research (PyTorch, vLLM, AMD ROCm, HAIM clinical embeddings) to solve performance, cost, and reliability bottlenecks in modern software.
Currently open for Software Engineering, AI/ML Engineering, and Systems Architecture opportunities.
GemmaRoute — 4-Layer Hybrid AI Routing Engine
- Core Stack: Python, Next.js 16, TypeScript, AMD ROCm, Fireworks AI
- Architecture & Impact: Built an intelligent 4-layer AI query router that cuts LLM API costs by up to 80%. Routes routine queries through local Gemma 2B running on AMD ROCm for free execution, escalating to cloud models (Fireworks AI) only when quality judges demand higher reasoning.
- Repository: jezreal-dev/gemmaroute | Frontend Dashboard
Themis — Autonomous AI Legal & Risk Analysis Platform
- Core Stack: Python, LangGraph, vLLM, AMD Radeon PRO W7900D (ROCm 7.2.1)
- Architecture & Impact: Architected a multi-agent legal risk analysis engine utilizing vLLM speculative decoding on AMD hardware. Implemented parallel LangGraph agents to parse complex compliance documents and identify regulatory risks.
- Repository: jezreal-dev/themis
RSNA Knee Abnormality Detection — Multimodal Clinical AI
- Core Stack: Python, PyTorch, HAIM Embeddings, Computer Vision, DICOM
- Architecture & Impact: Developed a multimodal medical AI pipeline combining HAIM (Holistic AI in Medicine) clinical embeddings with 3D MRI volume processing to detect structural knee abnormalities and assist diagnostic workflows.
- Repository: jezreal-dev/rsna-knee-haim-multimodal
VoltIQ — Real-Time AI-Powered EV Fleet Charging Engine
- Core Stack: Go, AWS Serverless (Lambda, DynamoDB, TimeStream), Amazon Bedrock, MQTT
- Architecture & Impact: Engineered an IoT telematics pipeline ingesting high-frequency battery telemetry (voltage, temperature, SoC, SoH) from fleet electric vehicles via MQTT. Integrated LLM agentic workflows to optimize charging schedules dynamically, reducing peak-demand grid strain in Lagos, Nigeria.
- Repository: jezreal-dev/VoltIQ
ropa-sci — Cyber-Neon Multiplayer TUI Gaming Platform
- Core Stack: Go, WebSockets, Bubble Tea, Lipgloss, P2P Networking
- Architecture & Impact: Built a concurrent CLI gaming engine featuring an interactive cyber-neon terminal user interface (TUI). Designed a predictive game-theory AI engine and thread-safe custom WebSocket server for real-time multiplayer states.
- Repository: jezreal-dev/ropa-sci
CleanPadi — Inclusive Multi-Channel Waste Management Platform
- Core Stack: TypeScript, HTML/CSS, WhatsApp API, USSD Gateway
- Architecture & Impact: Developed a multi-channel platform connecting households to verified waste collectors in Ilorin, Kwara State. Built for Hack4Dev to improve urban sanitation accessibility.
- Repository: jezreal-dev/CleanPadi
| Domain | Primary Focus & Verified Codebase Stack |
|---|---|
| Languages & Runtimes | Go (Golang) [High-concurrency backends, goroutines, channels, low-latency I/O] • Python [AI/ML modeling, PyTorch, medical computer vision] • TypeScript [Next.js 16 web dashboards] |
| AI & Machine Learning | PyTorch, HAIM clinical embeddings, local LLM routing, LangGraph multi-agent systems, Amazon Bedrock, Gemini API |
| Distributed Systems & Cloud | AWS Serverless (IoT Core, Lambda, DynamoDB, TimeStream), P2P WebSockets, Event-Driven Architecture (EDA), Docker containerization |
| Cybersecurity & Defense | OWASP vulnerability benchmarks, cryptographic protocol implementations, network sandboxing, secure code reviews |