San Francisco Bay Area
Software Engineer and Computer Science Senior at Arizona State University with a 3.81 GPA, graduating December 2026. I build full-stack AI products that ship to real users. Currently, I’m a Founding Software Engineer at STILED AI, where I built the platform from scratch across the React/TypeScript frontend, Node.js backend, Chrome extension, event-driven iMessage/SMS pipeline, GCP deployment, and Supabase database layer. STILED has 300+ beta users and has been featured in Business of Fashion. I’m also a Software Engineer at the AWS Cloud Innovation Center at ASU, where I build production RAG and agentic AI systems on AWS for partner organizations including Project ECHO and Cintana Education. Previously, I interned at AGCO Corporation, maintaining enterprise Salesforce systems for a 3,000+ dealer network. Outside of that, I’m building Trace, a memory and rule layer for AI coding agents. Trace is used in production by 4 companies, onboarded 15 developers in the first week, and won the VillageHacks ’26 Clean Track. My work spans full-stack engineering, AI systems, RAG pipelines, cloud infrastructure, developer tools, and agentic AI. Core technologies: TypeScript, JavaScript, Python, Java, C/C++, Go, Kotlin, SQL, React, Next.js, React Native, Node.js, Express, Flask, Django, PostgreSQL, Supabase, MongoDB, DynamoDB, Firebase, AWS, GCP, Azure, Docker, Kubernetes, GitHub Actions, LangChain, PyTorch, TensorFlow, HuggingFace, FAISS, ChromaDB, pgvector, Claude API, OpenAI API, and Gemini API. Open to full-time Software Engineer and SWE Intern roles. Relocating to San Francisco. Reach me at [email protected]
Project ECHO — RAG Chatbot Led a 3-person team building a retrieval-augmented generation (RAG) chatbot for Project ECHO, a global healthcare NGO scaling expert medical knowledge to frontline workers. Architected to serve 250K+ healthcare workers across 38 countries. Deployed on Amazon EKS Fargate with API Gateway and ALB. Orchestration via AWS Strands SDK. Built the full React/TypeScript frontend using Shadcn/UI. Among the first teams to use AWS Bedrock vector storage in preview — ingestion pipeline runs through S3-triggered Lambda, Bedrock Data Automation, and Titan Multimodal Embeddings. 90% answer accuracy with Bedrock Nova Lite. Implemented PII redaction and sanitization for healthcare compliance. Stack: Python, TypeScript, React, AWS Bedrock, Amazon EKS, AWS Lambda, S3, API Gateway, Titan Embeddings, vector search, RAG, LLM orchestration. Cintana Education — Agentic AI Admissions System Architected a multilingual agentic AI system on Bedrock AgentCore for Cintana Education, a global higher-ed network connecting students across 30+ universities in 28 countries. Orchestrates Claude Sonnet, containerized via Docker/ECR, infrastructure in AWS CDK. Built the React/TypeScript frontend end-to-end. Event-driven WhatsApp pipeline using SQS, Lambda, and Twilio with DynamoDB session tracking. Integrated Salesforce CRM and AWS Translate for multilingual RAG. Designed an extensible tool-use architecture where new agent capabilities plug in without modifying core logic. Stack: Python, TypeScript, React, AWS Bedrock, AWS CDK, Docker, ECR, SQS, Lambda, DynamoDB, Salesforce API, Twilio, Claude API, agentic AI, tool use, multilingual NLP, RAG.
Founding engineer on a 2-person team. Built the entire full-stack platform from scratch: React and TypeScript frontend, Node.js and TypeScript backend, Chrome extension, deployed on Google Cloud Platform with Supabase (PostgreSQL), CI/CD through GitHub Actions. Built an event-driven async iMessage/SMS pipeline in Node.js — 11 intent classification flows, a 7-state conversation engine, per-phone session isolation, zero dropped messages. The system captures purchase intent within a 2-minute styling window, which is the kind of real-time distributed systems problem I love working on. Integrated Google Gemini Flash 2.5 to power an agentic outfit recommendation engine with AI image generation. Built automated web scrapers with measurement normalization across 10+ retail websites, handling dynamic content and schema variations. Reduced AI inference latency from 40 seconds to 3 seconds through async image processing and model optimization (Gemini Flash → Flash-Lite swap, 76% per-call cost reduction). 300+ beta users validated the product: 89% found suggestions useful, 92.5% said they'd use it while shopping. Featured in Business of Fashion and AllThingsFashionTech. Live at thestiled.com. Stack: TypeScript, React, Node.js, Express, Google Cloud Platform, Supabase, PostgreSQL, Gemini API, Chrome Extensions, GitHub Actions, REST APIs, event-driven architecture.
Building an intent-aware memory and rule layer for AI coding agents like Claude Code and Codex. Used in production by 4 companies, onboarded 15 developers in the first week. Extracts six typed memory categories via GPT-4o Mini, retrieved through pgvector cosine similarity and Postgres full-text search over a Voyage-3 embedding space. Built an automated truth-maintenance layer with 100% contradiction recall and zero dangerous confusions — dead ends are stored as permanent hard constraints and never retried across sessions. Team-wide propagation via Supabase-backed shared memory with per-user attribution. 52% token reduction and 24% latency reduction vs cold sessions. Won VillageHacks '26 Clean Track. Accepted into DevLabs Momentum Cohort. Received founding-engineer offer from tryclean.ai.
• Empower peers by demonstrating the use of Adobe Express to enhance academic and personal projects. • Plan and execute 2+ Adobe workshops each year teaching students how to best utilize Adobe applications. • Collaborate with faculty, staff, and student organizations to highlight availability of Adobe products on-campus. • Promote Adobe through content creation and advertising through personal social media channels. • Connect with a global community of Adobe Student Ambassadors to exchange ideas and best practices.