AI engineering · backend systems · full-stack development
I work across model internals and the applications around them: implementing learning algorithms, building inference services, and connecting APIs to usable interfaces. My repositories include both focused learning experiments and larger application projects.
| Project | Engineering focus |
|---|---|
| LLM Inference Runtime | Language-model inference systems |
| EasyBid | Auction bidding, transactional updates, account recovery, and role-based access |
| ResolveAI Support Platform | Support-ticket triage and review, with a no-key demo and an optional AI workflow |
| AI Inference over gRPC | Four RPC patterns, bounded streaming requests, and a replaceable inference adapter |
| GPT from Scratch | A character-level decoder Transformer implemented in PyTorch |
| Amharic Word Embeddings | Skip-gram, manual gradients, numerical checks, and reproducible experiments |
ResolveAI uses fictional support data. The model projects are small experiments intended to make the underlying algorithms inspectable.
Models and data: Python, PyTorch, NumPy, scikit-learn
Services: FastAPI, Node.js, Express, gRPC, PostgreSQL, MongoDB
Interfaces and delivery: React, TypeScript, Docker, GitHub Actions, Vercel
I keep setup instructions, tests, and known limitations close to the code. Coursework and early exercises remain separate from the projects featured above.