Tempe, Arizona, United States
• Configured ephemeral Kubernetes preview environments for the portal monorepo using HAL and GitHub Actions — per-PR isolated deployments with automatic teardown, eliminating orphaned pod budget flooding in the non-prod cluster 97% and improving chat latency 23–30% • Shipped production intent journeys for a LLM-powered chat agent on a hard deadline; replaced non-deterministic LLM routing with deterministic branching, wired to 100+ FormAssembly operational forms, unblocking the sunset of a legacy support portal • Resolved production bugs across the stack by tracing failures to root cause: fixed a NestJS BFF throwing HTTP 500 on all up stream errors (2.33% failure rate), caught stale Angular state leaving partners stuck post rate-lock, and traced a false compliance failure through 3 service layers to a single missing cap check • Migrated a 122-suite Angular test suite Jest to Vitest, reducing CI runtime from 215s to 34s (84% faster) by resolving ESM/CommonJS incompatibilities • Replaced a stateless AWS Lambda with a Python async pipeline using semaphore-gated concurrency and Unix socket streaming resolved cascading failures under load, cutting Bedrock API calls
• Build production backend services in Python/FastAPI and TypeScript for Keysha, a voice-first AI assistant with chat capabilities, powering natural-language task execution, reminders, and multi-calendar workflows • Engineer LiveKit-based real-time voice interaction pipelines for streaming, low-latency conversations, supporting interruptible turn-taking and contributing to Keysha’s sub-50ms response-time product experience
• Served as a Computer Science Grader at Arizona State University for CSE 475 and CSE 463 courses. • Developed an AI-powered content detection platform using FastAPI and React, enhancing grading workflows. • Improved evaluation accuracy by 40% through the integration of GPT-based APIs. • Designed a scalable backend for document ingestion and structured reporting across multiple course sections.
• Built and scaled a centralized IoT-driven digital advertising platform for managing distributed kiosks across multiple cities, replacing manual ad deployment workflows with a self-serve client dashboard. • Designed and implemented RESTful backend services using C# ASP .Net , Express.js and FastAPI, leveraging object-oriented design principles to support ad ingestion, scheduling, authentication, and analytics with sub-120ms API latency under production load. • Developed near real-time analytics and scheduling pipelines, processing telemetry from IoT devices and persisting data across PostgreSQL and MongoDB, with Redis caching, reducing dashboard latency by ~33%. • Containerized backend services using Docker, published images to Amazon ECR, and deployed workloads across AWS EC2 and Lambda, leveraging cloud platforms to ensure scalable and reliable operations.
• Collaborated with a 12-member global team supporting regulatory reporting for 2 private banking clients, automating 200+ functional and API tests using version control systems to improve defect detection by 25%, reduce manual effort by 43%, and ensure reliable backend data validation • Designed and implemented automated test frameworks in Java and Selenium, incorporating debugging strategies to validate backend API and UI flows across multiple banking platforms, resulting in faster release cycles and higher test reliability • Built and maintained TestNG-based suites integrated with Jenkins CI pipelines, enabling nightly regression runs and consistent deployment validation