Kuanyao Alex Huang

About Kuanyao Alex Huang

I am a licensed medical doctor, software engineer, and machine learning engineer. My personal experience includes caring patients at the first line, building software systems, and crafting machine learning models to solve the challenging applications of healthcare systems and the crawl engine.I received the computer science master degree from Columbia University in Dec. I am especially interested in the applications of advanced machine learning models on visual, natural language, and cross-modality tasks. During my previous internship at Google, I independently built the first transformer-based semisupervised label propagation service on crawl signals using C++ and Python for my team.I finished the 7+ year training of the medical doctor program at National Taiwan University, received the medical doctor degree, and passed the medical license exams in Taiwan in June, 2020. I am equipped with the essential training from nearly all of the departments such as internal medicine, surgery, gynecology, pediatric, dermatology.etc. I am especially interested in applying machine learning algorithm on medical challenges, including the gigabyte-sized pathology images or dermatology images.My personal goal is to improve human\'s lives with technology and make people trust technology. I am now searching for a full-time software engineer or machine learning position in the US to pave the way for my goal.

Experience

1. Software Engineer

   Google

   Apr 2023 — Present · New York, NY, US

   Search for Storage — Google’s full-text search on Spanner- Architected the GoogleSQL VECTOR data type and SFP4/SFP8 quantization, slashing storage costs by 4x–8x and reducing Retrieval-Augmented Generation latency by >2x- Implemented the hybrid search feature combining AI-powered semantic search (ANN) with traditional tokenized results via Reciprocal-Rank Fusion to maximize retrieval relevance- Optimized Full-Text Search internals, including algebraization and query plan generation; launched new tokenization features (TOKENIZE_ENUM)- Launched real-time Autocomplete for Spanner and ST-Spanner, supporting high-frequency, personalized suggestions for 4+ major product teams.Authorization and Index Optimization (Large-Scale Graph Partitioning)- Optimized Google Drive’s global sharing graph by clustering documents with similar sharing patterns into the same physical database tablets to minimize distributed query overhead by 10x- Launched the \"Affinity Rollout\" strategy to migrate documents between optimal placements from two graph snapshots over weeks; by batching moves based on original placement, reduced search tablet-fanout by 40% and prevented transient performance degradation- Refined the Iterative Greedy Matching algorithm with regularization, reducing average search placement fanout by 10x and drastically cutting query costs across the global sharing graph- Performed schema migrations for Google Drive’s sharing placement feature, enabling query isolation and facilitating the seamless migration of private Drive search to Spanner.Structured Search on Bigtable (ST-BTI)- Orchestrated a global migration ecosystem, building auto-migration infra and data-diffing pipelines that enabled schemas to dark-launch on ST-Spanner, saving 10% in annual compute units- Developed the VanillaST syntax parser and schema validation engine, enabling 100+ teams to translate user-facing queries (e.g, Gmail from:/ to:) into optimized syntax tree.

Education

• National Taiwan University

  Bachelor of Science - BS, Physics

  2012 — 2020

• Taipei Municipal Jianguo High School

  High School Diploma, General Studies

  2009 — 2012

• National Taiwan University

  Doctor of Medicine - MD, Medicine

  2012 — 2020

• Columbia University

  Master of Science, Computer Science