A RAG backend on AWS that turns a daily-living goal into grounded, step-by-step plans, each citing its source. Three measured eval rounds on CanPlan's step-generation service ended at 100% valid citations and a reading level cut from grade 7.2 to 6.0 — and showed the safety guardrail still over-flags (96% → 89% of plans).

Fine-tuned a small language model to read financial text as bullish, bearish, or neutral — F1 0.83 while training only 1.31% of params (a 3.4 MB adapter) — then wrapped it in a chat agent that pulls live headlines. On live news it scores 0.66, level with FinBERT, and a fixed pipeline matched the agent's reliability with 7× fewer tokens.

An e-commerce order backend split into small services that never oversell stock, even under load — 599 concurrent orders on 200 units settle to exactly 200. It sustains ~1,884 req/s at 56 ms p95, and a connection-pool tuning round cut p95 21% (140 → 110 ms).

Mapped the size/F1 Pareto front for a CRDNN speech-detection model from 0.435 MB down to 0.050 MB. The shippable point is 0.185 MB at F1 0.943, against a 0.959 baseline — squeezing to 0.050 MB is 8.7× smaller but costs 9 points of F1. Dynamic quantization barely moved it, because only 1.2% of the parameters were quantizable.

A book recommender that chews through 30 GB+ of public library data into a 50,000-book catalog, then fans queries out across 26 workers / 27 partitions — figuring out which steps actually benefit from going parallel.

A full-stack movie social network where people register, follow each other, and post reviews — then like and vote on both movies and reviews. A React + Redux front end talks to an Express / MongoDB API that handles sessions and the social graph, and proxies live movie data from the TMDB API.