Hi! I am Colin Zhou.....

Montreal | Kitchener | Cupertino | Seattle | Champaign | San Mateo | ?

My time in San Mateo:

Colin Zhou at Snoqualmie Falls, Washington State, April 2022
*Me at Snoqualmie Falls, Washington State in April 2022

I graduated from UIUC with a Masters in Computer Science, and obtained my bachelor's degree in Computer Science from Bellevue College. I was born in Canada, but grew up in the Bay Area, lived near Seattle, Washington for 6 years, and am now in San Mateo, California.

I am currently a software engineer at Kaliber Labs, helping integrate multimodal LLM's into consumer products in the healthcare space. Previously, I worked at Nuclearn.ai (nuclear power startup) as a full stack engineer, and did multiple internships in the civics, telecom, environmental, and educational sectors. I am broadly interested in integrating the latest AI into useful products and how state of the art machine learning inference systems serve tokens to millions of customers.

When I'm not sitting at the computer, you can find me solving Rubik's Cubes of various sizes, practicing a difficult piano piece and posting it on my Instagram, doing 1000 piece jigsaw puzzles with my sister, or taking long walks around my neighborhood. You can also find me writing technical articles at my blog.

Education

  1. University of Illinois Urbana-Champaign

    Masters in Computer Science

    Activities: CS 411 Database Project Team Lead

    Relevant classes: User-Interface Design, Databases, Advanced Data Management, Human-Computer Interaction, Artificial Intelligence, Distributed Systems, ML + Data Systems

  2. Bellevue College

    B.S. in Computer Science

    Activities: CS Student Advisory Board Treasurer

    Relevant classes: Data Structures, Algorithms, Cloud Computing, Operating Systems, Software Engineering, Computer Architecture, Internet of Things, Computer Networks, Linear Algebra, Senior Capstone I, II, III, Discrete Math, Calculus III

  3. Cupertino High School

    High School Diploma

    Activities: Speech and Debate Club Member, Sophomore English Tutor

    Relevant classes: AP Computer Science A, AP Statistics, AP Physics C: Mechanics, AP Chemistry, AP United States History, AP Calculus AB

Experience

Full Stack Software Engineer

Movendi

A Kaliber Labs company, formerly Vitesse

08/2026 - present • San Mateo, CA, United States (Onsite)

  • Used TypeScript and Go to connect the historian microservice to the frontend, allowing users to query and download historical data from PLC controllers within a specified time interval and visualize the changes.
  • Implemented organizational RBAC controls over the integrations side of the platform
  • Used Go to implement functionality allowing users to replay SCXML statecharts with data from historian, helping factory owners test an SCXML diagram before deploying to edge devices on the factory floor.

Software Engineer

Kaliber Labs Inc.

01/2026 - 08/2026 • San Mateo, CA, United States (Onsite)

  • Rearchitected the Conversation Pipeline demo to be more modular and extendable to various medical skills, such as Speech Therapy / Patient Intake using FastAPI and React. Successfully orchestrated multiple model operations within one system (Qwen, GPT).
  • Leveraged the Conversation System to create a data platform deployed on AWS for subject matter experts (SME's) situated around the US to verify and collect data on doctor-like responses from the in-house medical AI.

Full Stack Software Engineer

Nuclearn

12/2024 - 02/2025 • Phoenix, Arizona, United States (Remote)

  • Utilized FastAPI/SQLAlchemy to develop backend functionality for retrieving up to 25 most similar records in a dataset using 500+ embeddings generated through natural language techniques.
  • Architected a back-end workflow with new SQLAlchemy datamodels to migrate 100+ YAML configurations across five config files into PostgreSQL, enabling dynamic updates via a React frontend, eliminating the need for Nuclearn's developers to restart the web backend during development if such configurations were changed.
  • Optimized markdown processing for artifacts displayed across 10+ tool calls in AtomAssist, Nuclearn's proprietary AI chatbot.
  • Implemented role-based access control with error handling for consumer, admin, and contributor roles, ensuring secure access to 10+ unique web app features on the frontend like Cap AI, AtomAssist, Capitalizer, and Cluster Analytics

Open Source Software Engineer

PolicyEngine

03/2024 - present • Issaquah, Washington, United States (Remote)

  • Worked with 3 contributors on PolicyEngine's web app project on GitHub (over 600 visits monthly).
  • Assisted in migrating the front end component library to the newest version of Ant Design. Streamlined 150 Jest unit tests by refactoring 5 metadata API calls to occur in the beginning of testing runs.

Software Engineer Intern

EcoLogical

09/2023 - 12/2023 • Champaign, Illinois, United States (Remote)

  • Using Ant Design components to improve the front-end experience of the flagship web app for small businesses to track their carbon footprint.
  • Set up 10 mutations and queries to AWS Amplify backend for the user signup workflow.

Software Engineer Intern

IOIntelligence

07/2022 - 09/2022 • Bellevue, Washington, United States

  • Developed a 5 page application dashboard that displays networking analytics from company's wireless devices worldwide.
  • Used Chakra UI, React for frontend and Express.js and MySQL for the backend.

Software Engineer Intern

Molloy Moreno LLC

06/2021 - 09/2021 • Issaquah, Washington, United States

  • Developed a serverless web app for New York school teachers to track the academic progress of at-risk students.
  • Made final design choices for UI components.
  • Used React to implement UI components on the frontend and AWS for storing media on the backend.

Projects

LLM God

A desktop application that allows you to query 4 LLM's at once for a single input. Supports Claude, ChatGPT, and Gemini and more! As someone who uses LLM's a lot in my day to day life, I wanted an easy and convenient way to ask multiple LLM's at once, and it resulted in this project. It achieved over 270 stars on GitHub.

  • HTML
  • CSS
  • JavaScript
  • Node.js
  • Electron.js

nanoGPT Inference Optimizations

Screenshot of nanoGPT Inference Optimizations

Implementing 19 industry level ML inference optimizations on Andrej Karpathy's nanoGPT code, including KV Caching, Chunked Prefill, Speculative Decoding, Paged Attention, Continuous batching, etc. Benchmarked all these strategies against a baseline and also created an interactive frontend / backend that demonstrates the main differences and performance gains.

  • Python
  • PyTorch
  • FastAPI
  • React / Vite

StreetFoodLove

Screenshot of StreetFoodLove

For my senior year capstone project in undergraduate, I worked with three others to create a Yelp-like app that will empower and enable food truck vendors to more easily attract customers and run their business. I worked with mostly Typescript and React.js on the front end for developing the UI and used Adobe XD for the initial mockups. On the backend, I helped with writing SQL queries and making sure the AWS services were linked up seamlessly with the frontend.

  • HTML/CSS
  • TypeScript
  • React.js
  • Semantic UI
  • AWS
  • Redux
  • SQL

Project Elpis App

Screenshot of Project Elpis App

In my first internship, I helped create an educational web app that helps at risk high school students track their future career goals. I utilized various front end technologies like React.js along with backend tools like AWS for hosting. After the internship was completed, this app is now in beta testing in a high school of a couple hundred students in New York City.

  • HTML/CSS
  • TypeScript
  • React.js
  • Semantic UI
  • AWS
  • Redux

React Backend-Benchmarks

Screenshot of React Backend-Benchmarks

A web application with a React frontend and three backends written in different languages that are all connected to a common MySQL database. Users will be able to send repetitive queries to any backend that they choose and see how fast the server responds after the completed operation.

  • TypeScript
  • React.js
  • Node.js
  • MySQL
  • Python / Flask
  • Golang

Caption Image Classifier

A full stack web application that classifies images and generates captions; using my own pretrained PyTorch model based on the CIFAR-100 dataset and leveraging HuggingFace's API to do image captioning using WebSockets.

  • React
  • PyTorch
  • FastAPI
  • HuggingFace
  • WebSockets

Show Full Project Archive

Skills

Frontend

  • HTML
  • CSS
  • TypeScript
  • React
  • GraphQL
  • Figma
  • AdobeXD

Backend

  • Python
  • Apache Parquet
  • Go
  • Node.js
  • MongoDB
  • MySQL
  • PostgreSQL
  • FastAPI
  • MapReduce
  • Docker

Tools

  • Git
  • Jira
  • VSCode
  • LaTeX
  • Markdown
  • YAML
  • Linux
  • Windows
  • Prompt Engineering
  • AI Agents
  • NPM
  • Yarn
  • SCXML

Classwork

Distributed System

Fall 2023 · UIUC
  • Python
  • Linux
  • Hadoop

Web Programming

Winter 2022 · Bellevue
  • HTML/CSS
  • JS
  • React

Internet of Things

Winter 2022 · Bellevue
  • C++
  • C

Operating Systems

Fall 2021 · Bellevue
  • C

Data Structures / Databases

Fall 2020 · Bellevue
  • Java
  • C++

Programming Languages

Winter 2021 · Bellevue
  • Java
  • Python

Fundamentals of CS II

Winter 2020 · Bellevue
  • Java

Artificial Intelligence

Spring 2023 · UIUC
  • Python

Cloud Computing

Spring 2022 · Bellevue
  • C#

Spring 2021 Bellevue Work

Spring 2021 · Bellevue
  • Java

Hobbies

Colin's Instagram profile, colin.pianistColin's TikTok profile, colin.pianist

I started learning piano when I was 8 and took lessons for almost 10 years. I passed the Certificate of Merit Level 10 exam and also went to multiple piano competitions on top of yearly recitals. After quitting private lessons in high school due to work, I began to play more as a hobby, and started uploading footage of myself trying out challenging pieces on my Instagram and TikTok. I play mostly classical composers like Chopin, but also a large variety of other genres too!

I am a fan of Rubik's Cube speedsolving and have participated in multiple official competitions since 2014. My best time in a competition was 12.92 seconds. Here is a video of myself solving a 3x3 cube in under 14 seconds at a Bay Area event in 2015.

🤖 For AI Agents

This portfolio is agent-optimized. Download machine-readable files below.

llms.txtFull-site index & instructions for AI agentsresume.mdFull resume in clean Markdown formatprojects.jsonAll projects as structured JSON

MCP Server Available

Query this portfolio programmatically via the Model Context Protocol at my-portfolio-mcp.colinzhou.workers.dev

Contact: colizu2020@gmail.com · @ColinZhouUSA · LinkedIn