Hello, I'm Alvin
AI Engineer · Developer Tools · Full-Stack
Building LLM agent systems, developer tools, and full-stack AI applications. Scroll for selected work, or jump to Projects.
Alvin Xu
Software engineer focused on AI agents, retrieval, and developer tools.
Education
Technical toolkit
- CiteLoom: Plan-Execute agents, two-stage retrieval, and an MCP inspection server.
- NeuralLens: interpretability tools for robot models, with hooks into the simulation loop.
- Evaluate retrieval with NDCG and MRR, and measure cost and variance across agent runs.
- Record design decisions and verify changes with end-to-end browser tests.
Selected work
All projectsCiteLoom — Independent AI Product
Tauri-based AI research canvas combining a Plan-Execute LLM agent loop and a two-stage BGE + RRF retrieval pipeline. 29 ADRs, an MCP-compatible inspection/automation server (163 endpoints across 6 namespaces), and a 4-layer Rust DB architecture.
NeuralLens — Mechanistic Interpretability Toolkit
Open-source Python library (in development) for inspecting and steering robot foundation models — VLA models, world models, RL policies — with activation inspection, feature steering, ablation, and live simulation-loop hooks.
Investment Platform — Full-stack Web & Android
Angular, Express, and MongoDB app with market data from Finnhub and Polygon. The web demo runs on Vercel and includes a simulated portfolio.
6-DOF Robotic Arm — Motion Planning & Control
Motion planning + localization framework (RRT trajectory planning, MCL with LiDAR) and grasping (TSR-based IK sampling + Jacobian methods), validated on both RViz simulation and a real robotic arm.
End-to-End NLP Pipeline (Classical ML → Transformers)
Sentiment, POS, and NER pipelines on Amazon Reviews and CoNLL-2003. Benchmarked classical ML (SVM, HMM, LR) against neural models (Word2Vec, GloVe, BiLSTM, GRU, Transformer); BiLSTM+GloVe reached 88% F1 on NER.