I'm a Computer Science student at Arizona State University building AI agents and production software at the boundary between customer workflows and engineering.
I work from the operating problem outward: understand how people actually work, define the data and integration requirements, build the system, test failure modes, and stay through rollout and iteration.
- AI agent deployments: RAG, vector retrieval, structured tool calls, REST APIs, deterministic fallbacks, and human escalation for service-business workflows.
- Production backend systems: TypeScript, Python, PostgreSQL, workflow state, concurrency control, CI, E2E testing, monitoring, and recovery.
- Applied ML evaluation: leakage-resistant experimental design, pretrained-model fine-tuning, variance analysis, and honest reporting of limitations.
A production-oriented SIS built with NestJS, Next.js, PostgreSQL, Prisma, and TypeScript. It includes concurrency-safe enrollment, waitlists, prerequisites, RBAC, API and browser E2E tests, monitoring, and backup-restore drills.
A live Next.js/Supabase platform for reservations, Stripe payments, donations, and back-office operations. Reliability work includes server-side pricing, idempotent checkout, signed webhooks, rate limiting, reconciliation, and operational risk monitoring.
Subject-disjoint experiments comparing EEGNet with the pretrained LaBraM foundation model on PhysioNet EEG data. The project documents both the performance gain from matched preprocessing and the limits of a high-variance, single-split result.
Discover → Scope → Build → Evaluate → Roll out → Observe → Iterate
Python TypeScript JavaScript SQL PostgreSQL Next.js NestJS Prisma Supabase Docker PyTorch RAG Tool Calling REST APIs