Full-stack developer working mainly in Java / Spring Boot, with C# / .NET and React / Angular on the side. These days I build LLM features the way backend systems should be built: measured with evals, protected against prompt injection, and limited on cost.
| Project | What it is | Stack |
|---|---|---|
| token-allowance | Spring Boot starter that enforces per-user, per-tenant and per-session token and dollar budgets on Spring AI calls, and stops runaway tool loops. Demo runs on a local model, no API key needed. | Java, Spring AI |
| retrieval-chatbot | RAG chatbot with hybrid search (BM25 + vectors), a cross-encoder reranker, and three eval suites: retrieval recall, answer accuracy and prompt-injection attack rate. | Python, sentence-transformers, Ollama |
| tradehub | Mini trading platform: JWT auth, buy and sell orders against simulated live prices, portfolio and profit/loss tracking. | C#, ASP.NET Core 8, EF Core, React |
| taskflow | Task management app with a REST API, JWT auth and Swagger docs (web client). | Java 21, Spring Boot, PostgreSQL, TypeScript |
| tinyurl-generator | URL shortener with Base62 codes, click analytics and a scheduled cleanup of expired links. | Java 21, Spring Boot, PostgreSQL, React |
Backend: Java, Spring Boot, Spring AI, C#, ASP.NET Core, Python
Frontend: React, Angular, TypeScript, Tailwind
Data and infra: PostgreSQL, Docker, Ollama
