Hi, I'm
Michael Ku Jr
Software & AI Engineer
I build agents, RAG pipelines, and full-stack tools — ideas taken all the way to done.
Hi, I'm Michael — an Electrical & Computer Engineering student at Olin College who loves building at the intersection of AI and software. I care about shipping things that work: clean code, real users, and ideas taken all the way to done.
Projects
Things I've built.
Universal Document Intelligence Platform
A configurable RAG-and-reasoning system for finance, legal, and healthcare docs — hybrid retrieval, cross-encoder reranking, and a rigorous eval harness that benchmarks every change against a vanilla-LLM baseline.
- Python
- Claude Sonnet
- Qdrant
- FastAPI
Personal Learning Agent
A multi-agent RAG system that ingests AI/ML content, builds a hybrid concept-and-tool knowledge base in Obsidian, and answers adaptively — with video timestamps and watch-tracking so nothing gets re-taught.
- Python
- LangGraph
- LangChain
- Chroma
AI Agent Library
A modular Python framework for assembling LLM agents — chain tools, APIs, and data sources into reusable workflows without rebuilding orchestration each time.
- Python
- LangChain
- LangGraph
- FastAPI
AI Fridge Camera
A computer-vision system that auto-logs what enters and leaves your fridge, tracks freshness, and suggests recipes before food expires — full-stack AI on a Raspberry Pi.
- Python
- OpenAI Vision
- OpenCV
- Raspberry Pi
LLM Eval Harness
An open-source framework for measuring whether an LLM or RAG system actually works — LLM-as-judge scoring, retrieval metrics, and regression tracking that turns "seems better" into a number you can trust.
- Python
- pytest
- Claude
- LangSmith
LoRA Fine-Tuning Lab
Fine-tuning a small open-weight model with LoRA on a focused domain task, benchmarked before-and-after against the base model — proving out training, not just prompting.
- Python
- PyTorch
- Hugging Face
- PEFT / LoRA
Scribblz — Wall-Drawing Robot
An autonomous robot that climbs vertical glass and draws on it, using suction adhesion, holonomic motion, and computer-vision docking. I led the electrical design.
- Custom PCB
- Python
- Computer Vision
- Robotics
3D Scanner
An Arduino-driven pan-tilt scanner that sweeps an IR sensor across an object and reconstructs its shape in MATLAB — calibrated optics, clean serial pipeline, real results.
- Arduino
- C++
- MATLAB
- 3D Printing
Balloons TD
A full tower-defense game built from scratch in Python and Pygame — pathfinding, collision detection, upgradeable towers, and wave economy, all at a smooth 60 FPS.
- Python
- Pygame
- OOP
- Game Dev
Experience
Where I've worked & led.
- May 2026 — Present
Liberty Mutual
Full-Stack Engineering Intern · Global Risk Solutions
On the TechStart team in Global Risk Solutions, developing a new cyber-risk platform end-to-end — backend, API, and UI — to replace legacy systems and give commercial teams faster, self-serve implementation.
- Built the service and API layer on a Directus backend, defining OpenAPI/Swagger contracts to standardize data flow between database and application.
- Engineered the web application UI in Angular and TypeScript, turning platform requirements into production components.
- Introduced MCP-based tooling and modernized legacy web packages, building automated feedback loops that accelerated team delivery.
- 2025 — Present
Limitless
Co-Founder & CTO
Co-founder and CTO of an early-stage startup — architecting and building the full platform end-to-end (Next.js, TypeScript, Tailwind, Supabase, Vercel) and owning engineering from zero to launch.
- Established the entire engineering foundation from zero: environment structure, data model, and CI/CD pipeline.
- Developed proprietary AI models to interpret and evaluate idea quality, powering the platform's core discovery engine.
- Own architecture, infrastructure, and deployment through to launch.
- May 2025 — Aug 2025
InnovationForce
Software Engineering Intern
Built and deployed Retrieval-Augmented Generation (RAG) agents and Crawl4AI-based web crawlers for a SaaS innovation platform — automating research workflows that saved teams 100+ hours.
- Built and deployed RAG agents that improved user self-service adoption and efficiency.
- Designed web crawlers with Crawl4AI to automate market research and reduce analyst costs.
- Developed a React-based GUI that simplified database scraping for nontechnical users and improved client onboarding.
- 2023 — Present
Ascend Pan-Asian Leadership
President & Founder, Olin Chapter
Founded and lead Olin's first Pan-Asian leadership chapter, part of North America's largest Pan-Asian professional organization — and won the $10,000 Ascend Legacy Scholarship.
- Created consulting opportunities that placed members in internships at partner companies.
- Run workshops, mentorship, and networking programs connected to Ascend's national network.
- 2023 — 2025
Olin Baja
Electrical Subteam Lead
Leading the electrical team toward a data-driven approach — building sensor modules (hall-effect, GPS, IMU) for real-time diagnostics on an SAE Baja off-road race vehicle.
- 2023 — Present
Ethos Angel Investment
Research Analyst
Researched portfolio companies and market trends for a Canadian angel fund, compiling the semi-annual reports that kept investors informed on performance and positioning.
Contact
Let's build something.
I'm always open to new ideas, internships, and collaborations. The fastest way to reach me is email.
michael.ku@outlook.com