open to internships & new grad for summer / fall 2026

Harmit Patel

CS student at University of Windsor with an AI specialization and a minor in Mathematics, graduating April 2026. I build full-stack web apps, train ML models, and care about systems design.

Currently Teaching Assistant for COMP 2140 (Compiler Design) at UWindsor. Previously completed Computer Engineering Technician studies at Niagara College, giving me a hardware, embedded systems, networking, and IoT foundation alongside software engineering.

Based in Canada 🍁🍂

skills

languages

Python TypeScript JavaScript Java C C++ C# MATLAB BASH SQL Go Rust Assembly

frontend

React Angular Next.js HTML5 CSS3 Tailwind SCSS/SASS Bootstrap Redux Webpack NPM/YARN/BUN

backend & databases

Node.js Express NestJS .NET Core GraphQL SQL Socket.io Firebase Convex Clerk Postgres Prisma Cassandra MongoDB Redis SQLite

technologies & architectures

PyTorch RAG Langgraph ChromaDB Langchain Ollama CUDA LLM Architectures REST APIs Serverless Docker Kubernetes Git GitHub Actions ROS2 AWS Azure Vercel Kafka

experience

Teaching Assistant — COMP 2310: Theoretical Foundations of Computer Science University of Windsor · Supervisor: Dr. Peter Tsin

Jan 2025 – Aug 2025

Supported students with automata, formal languages, proof techniques, induction, recursion, logic, and theory-heavy problem solving. Assisted with grading, office hours, and clear step-by-step explanations for abstract CS concepts.

Automata Theory Formal Proofs Induction Student Mentoring

Teaching Assistant — COMP 2140: Languages, Grammars and Translators University of Windsor · Supervisor: Dr. Jianguo Lu

Sep 2025 – Apr 2026

Assisted students with grammars, parsing, compiler concepts, translation systems, and programming-language fundamentals. Helped students connect theory with implementation-level reasoning through lab support, grading, and office-hour explanations.

Compiler Concepts Grammars Parsing Course Support

projects

check all projects I commit · github ↗

⚙

LeetCloned — coding practice platform LeetCode-inspired platform with curated algorithm problems, a code editor workflow, Firebase authentication, Firestore progress tracking, and a responsive Next.js interface. The project is deployed , solve some problems over there to make your problem solving skills better.

GitHub Repo Live on Vercel Next.js 14 TypeScript Firebase Tailwind

read codebase ↗

🧠

AlgoMentor — Socratic DSA tutor Local tutoring app for LeetCode-style problems. It uses a RAG pipeline with ChromaDB, sentence embeddings, Gradio, Ollama, and Llama 3.2 to give guided hints instead of direct answers.

Python RAG Ollama Llama 3.2 ChromaDB Gradio

read codebase ↗

🤖

OpenVLA Prompt Sensitivity Research Research project studying how robot-action predictions change when natural-language instructions are paraphrased. Includes prompt variants, failure analysis, CSV outputs, and paper-ready experiment plots.

OpenVLA Robotics Python Vision-Language-Action Failure Analysis

read codebase ↗

💬

ChatiFly — real-time communication platform Modern chat application with secure authentication, private and group messaging, multimedia sharing, video calls, real-time status, and AI-powered chat/image features.

TypeScript React Convex Clerk ZegoCloud OpenAI

read codebase ↗

👁

Part-Aware Superpixel GNN for Robotics Graph-based household object-recognition project using SLIC superpixels, region adjacency graphs, RGB-D features, GCN/GAT variants, and instance-level evaluation for robotic perception.

PyTorch Geometric GNN GCN GAT RGB-D Robotics

read codebase ↗

🔬

Superpixel Graphs on CIFAR-10 Object-recognition experiment that converts CIFAR-10 images into superpixel graphs and compares GraphSAGE, GCN, and GAT models to study graph-based image representations.

CIFAR-10 SLIC GraphSAGE GCN GAT Jupyter

read codebase ↗

📋

Version Control Line Tracking Tool Software-engineering tool for source-line tracking across versions using line mapping and similarity metrics for version comparison, bug localization, refactoring analysis, and change-impact review.

Java Algorithms Software Engineering Line Mapping Change Analysis

read codebase ↗

writing

Why teaching theory made me a better engineer A detailed reflection on how teaching theory changed the way I write code: proof structure, clean assumptions, edge cases, and why a good explanation is a stronger test than memorizing a solution.

Feb 2026

essay

Building a Git clone from scratch A codebase walkthrough of my version-control project: staging, snapshots, hashing, repository objects, line mapping, and how building a smaller Git-like tool made real Git feel less mysterious.

Jan 2026

coding

Things I believe — as a CS + AI student in 2026 A personal engineering manifesto about AI tools, fundamentals, public proof of work, readable systems, and the habits I think matter most for students trying to become real software engineers.

Dec 2025

thoughts

What LeetCloned taught me about product engineering A detailed project insight from LeetCloned covering authentication, problem data, editor UX, Vercel deployment, Firebase structure, and why finishing a product matters more than adding endless features.

Mar 2026

project insight

RAG is not magic: notes from AlgoMentor A longer BlogSpot-style breakdown of AlgoMentor: RAG design, vector search, local LLM constraints, prompt boundaries, and why AI tutors should guide problem solving without destroying learning.

Mar 2026

ai systems

Graph neural networks made images feel different A research-style note on GNN object recognition: converting images into superpixel graphs, comparing GCN/GAT/GraphSAGE, and learning how representation choices change model behavior.

Apr 2026

ml research

The boring stack usually wins A practical software-engineering essay on why stable tools, simple architecture, clean data models, monitoring, documentation, and readable code usually beat hype-driven technical decisions.

Apr 2026

tech philosophy

What today’s AI wave means for junior engineers A detailed career note about the 2026 AI wave: what changes for junior engineers, what does not change, and why debugging, taste, systems thinking, and shipped projects still matter.

May 2026

career notes

contact

"Driven by a desire to connect people and foster resilience, I view my work as a craft that blends creativity with impact, leaving the world a little brighter than I found it".