Software Engineer | AI Systems, Evaluation, Computer Vision
π B.S. Computer Science (math minor), Washington State University, May 2026 π Working on: AI-assisted review tools and evaluation of AI agents π± Learning: agent evaluation, MLOps, backend systems πΌ Looking for: entry-level software engineering roles building and testing AI systems
π Granny: open-source apple quality assessment (Python, YOLOv8, PyTorch, OpenCV) Contributed to a computer vision pipeline used by USDA-funded researchers. Redesigned starch detection with per-image adaptive thresholding (40% to 100% accuracy), consolidated multiprocessing into a shared base class (110+ duplicate lines removed), and set up the first CI, growing the test suite from 48 to 168 tests.
ποΈ SecureDraw: AI-assisted loan draw review (Python, Pydantic, PostgreSQL, Next.js, Claude) Claude reads draw-request PDFs into structured fields, then Python does exact-decimal math and runs 9 rule checks. A human decides every release, and decisions go into a hash-chained, tamper-evident audit log. Tested on 36 synthetic packages: 36/36 correct recommendations, 0 false approves. Independent project with synthetic data.
π¦οΈ AgWeatherNet AI Assistant (Python, Qwen2.5, Ollama, SQL, React) Team capstone: a self-hosted LLM assistant that answers plain-English questions over a 350+ station weather network and 4 databases. 29 question types go through a deterministic classifier before any model call.
β³ TeeTime (Next.js, TypeScript, Prisma, PostgreSQL) Rebuilt a live golf-tech product with OTP auth, revocable sessions, and database-level rate limiting for an AI-assisted search.
π£οΈ Fun fact: I speak English, Urdu, and Shina! And lived long-term on 3 different continents.


