Hi, I'm Bruce

Master of Science, Software Engineering (GPA: 3.5) | Mountain View

Coursework: Software Engineering, Foundation of Computer System, Functional Programming, AI in Industry, Gen AI in Practice, Machine Learning in Software Engineering

Bachelor of Science, Mathematics-Computer Science (GPA: 3.6)

Coursework: Software Engineering, Design & Analysis of Algorithms, Advanced Data Structures, Computer Systems, Parallel Computing, Computer Security, Online Database Applications, Recommendation Systems, Computer Vision

Led product strategy for an AI-driven call assistant across mortgage, healthcare, insurance, and government sectors, guiding 2+ end-to-end product cycles from discovery to launch.

• Designed and shipped an AI-powered dialogue flow system using FastAPI and a curated knowledge base, cutting average handling time by 35%.
• Redesigned the intent-matching and recommendation engine, migrating from Google Dialogflow to Transformer + ID with Gemini 2.0 and Vertex AI, reducing per-intent cost from $0.002 to $0.00031 (6× cheaper).
• Partnered with engineering and R&D to define technical requirements and manage agile sprints, reducing feature development time from 14 to 5 days.
• Ran A/B tests and analytics that improved AI model accuracy by 35% and increased sales conversion by 15%.
• Promoted from intern to full-time AI Product Developer in March 2025 after building a data-processing pipeline with Pandas and NumPy that improved preprocessing efficiency.

Built production-ready features for a web-based AI medical chatbot, focusing on performance, reliability, and patient experience.

• Developed a FastAPI + LangChain backend with Google Gemini and Google Cloud to accurately identify 50+ common diseases from patient queries.
• Implemented RAG-based knowledge retrieval on secure cloud infrastructure using Terraform and MySQL, reducing average response time from 2.5s to 1.5s and improving answer relevance for follow-up questions.
• Integrated Google Calendar API for appointment booking and enhanced UI/UX through iterative design reviews, enabling 50+ bookings during pilot testing.
• Led backend integration of the chatbot with LangServe and LangGraph using FastAPI and NestJS, streamlining deployment and improving overall chatbot performance.

• Built a LangGraph-based AI agent with a RAG pipeline to answer questions from 30 indexed academic papers.
• Developed a Streamlit web app for paper upload, multilingual querying, and context-grounded answers with references.
• Deployed containerized services including the AI agent, translation API, and vector database on Google Kubernetes Engine with auto-scaling.
• Implemented semantic retrieval using vector embeddings and compared indexing algorithms including HNSW, IVF_PQ, and DiskANN.

Built a Bosch-sponsored parking spot locator using vision-language models to semantically identify parking availability from visual sensor data.

• Implemented a VLMap-based pipeline combining RGB, depth, and pose data with CLIP embeddings for spatial reasoning.
• Developed and deployed a FastAPI backend on AWS to support real-time parking queries and semantic search.
• Optimized the semantic search pipeline, reducing parking availability query latency from 45s to 15s through improved embedding retrieval and backend query processing.

psl.fogx.link

Led full-stack development of an AI-powered banking chatbot using GPT-4o, Node.js, Flask, and React.

• Reduced customer service response time by 40%.
• Implemented NLP-based query handling with text and voice output, improving banking query resolution accuracy from 68% to 85%.
• Designed a dynamic real-time UI/UX that improved engagement and retention during pilot testing.

Built a blockchain-based medical data exchange platform using React.js, Express.js, MySQL, Web3.js, and JWT authentication to securely connect pet owners, clinics, and insurers.

• Implemented token-based on-chain record storage and role-specific dashboards for secure, permission-based access.
• Optimized backend APIs to cut data retrieval time from 3s to <1s, improving platform responsiveness.