Skip to content

Work

Backend systems, agents, and the evaluation harnesses behind them.

  • An agentic RAG system for macOS troubleshooting — a branching LangGraph, a domain knowledge base built from 9 scrapers, and a blind-judge evaluation harness that blocks its own deploys on quality regressions.

    · Solo — design, build, deploy, evaluate
    • langgraph
    • agentic-rag
    • evaluation
    • fastapi
    • python
    Live site
  • A coding agent built on LangGraph from scratch — a planner, a ReAct tool loop, and a network-disabled Docker sandbox — graded against 50 real Django and Sphinx issues from SWE-bench Verified Mini.

    · Solo — design, build, benchmark
    • langgraph
    • agents
    • evaluation
    • swe-bench
    • python
  • A Spring Boot REST API for personal finance tracking, with a security layer that returns proper 401/403 JSON instead of Spring's default HTML redirects, and JPA projections that push aggregation into SQL.

    · Solo
    • spring-boot
    • spring-security
    • java
    • jpa
    Source
  • A digital wallet decomposed into four Spring Boot services communicating over Kafka, with per-service data ownership and dedicated consumer groups for wallet-creation and transaction-update events.

    · Course project — solo
    • kafka
    • spring-boot
    • microservices
    • java
    Source
  • A hand-rolled reliability layer over raw UDP sockets in C — sequence numbering, acknowledgements, gap detection, and retransmission — built to understand what TCP does by implementing a subset of it.

    · Solo
    • c
    • networking
    • systems
    Source
  • The portfolio itself — an MDX content pipeline validated by Zod at build time, and a single shared WebGL canvas that's capability-gated and never blocks first paint.

    · Solo
    • nextjs
    • typescript
    • mdx
    • react-three-fiber
    Source