4th year CS @ UofT. Ex-SWE @ Veeva Systems. Building AI agents and RAG systems. Looking for Fall 2026 SWE internships.

Laith Alzoubi  • Peerlist

Used scraped Khan Academy data to build ML models (KNN, IRT, Neural Network) predicting student performance on diagnostic questions. Boosted runtime efficiency by 80% and reduced memory usage by 26% via PCA. Developed with Python and scikit-learn to explore personalized learning at scale.

🧠 AI Lab Inc. — Smarter learning through pre

Overview: The Contact List App is a web app that helps you keep track of your contacts. It's made to be easy to use so you can quickly find and manage people's names, emails, and phone numbers. Key Features: Simple to Use: Add, see, change, and remove contacts easily. Works on All Devices: The app looks good and works well on phones, tablets, and computers. Easy Interface: The design is clean and simple, making it easy to use. How to Start: If you want a better way to handle your contacts, try this app. Just download the code, set it up, and you're ready to go. It's great for keeping in touch with friends, family, or work contacts. With the Contact List App, managing your contacts is straightforward and efficient.

Contacts List Webapp

An AI-powered Android mobile application aimed to provide a predictive forecast on housing prices using React Native and scikit-learn.

EstiMap

Built a housing price prediction app using scikit-learn, with a React Native frontend and MongoDB + Express backend. Combines AI-driven forecasts with smooth UX for smarter real estate decisions.

🏡 EstiMap — Know the price before you knock.

Overview: Worked in a team to improve FortCraft, a turn-based game in JavaFX. We focused on making the game run smoothly and added new features. We used Git to manage our code and keep everything organized. Key Contributions: Game Development: Built FortCraft using JavaFX. Applied design patterns like Command, State, Singleton, and Memento to make the game logic better and easier to add to. User Interface: Made a user-friendly interface for players to interact with the game world, manage their items, and fight strategic battles against trolls. Accessibility Features: Added a font size slider and text-to-speech to make the game more accessible to everyone. Gameplay Features: Improved the game by adding different combat options like attack, heal, and escape. Also included dual color themes, a way to track health points during battles, and a settings window for players to customize their experience. FortCraft is more than just a game; it's a user-friendly, inclusive, and strategically engaging experience

FortCraft

Hermes is a real-time, AI-powered translation tool that keeps conversations flowing. Unlike sentence-by-sentence tools, it logs full dialogues for seamless cross-language communication. Built with React, Google Cloud APIs, Convex, and Cohere at Hack the North 2024.

🗣️ Hermes — Break the language barrier, not 

Built a Retrieval-Augmented Generation system with PDF ingestion, semantic search via pgvector, and cross-encoder reranking for precision retrieval. Generated cited answers grounded strictly in retrieved chunks to eliminate hallucination. Automated an eval harness scoring faithfulness, relevance, and context precision across 20 Q&A pairs.

🧠 RAG Pipeline — Precision retrieval, cited 

Built a real-time system ingesting live equity prices for 50 symbols via WebSockets, exposing current and historical data through REST APIs. Stored tick history in TimescaleDB hypertables for efficient time-series queries with a Redis caching layer for low-latency reads. Rendered live prices, historical tables, and system health metrics in a terminal dashboard.

📈 Real-Time Market Data Aggregator

Real-Time Market Data Aggregator Python · Fas

Built a real-time messaging app using the MERN stack and Socket.IO. Features include live chat, JWT-based auth, encrypted passwords, and a responsive React interface for seamless communication.

💬 RealTime Messenger — Chat fast, stay secur

Overview: RecipeHub is a web platform I developed using the MERN stack (MongoDB, Express.js, React, Node.js). It's designed to let users find, create, and share recipes, building a community of food enthusiasts. Key Features: Manage Your Recipes: Users can easily add and organize their own recipes, including ingredients, steps, and descriptions. Explore New Flavors: Thanks to Spoonacular API, users can find new recipes and get inspired to try different cuisines. Safe and Secure: We use JWT (JSON Web Tokens) for security, so user information is protected and only registered members can add to the site. Share and Inspire: Users are encouraged to share their culinary creations, helping others discover and try new dishes. Technologies Used: MongoDB Express.js React Node.js JWT for Authentication Spoonacular API RecipeHub is more than just a recipe site; it's a platform for culinary discovery and sharing.

RecipeHub Web App

Upload your resume and instantly get a personalized, step-by-step roadmap to your dream job. RoadMapr analyzes your experience, identifies skill gaps, and suggests roles, companies, and next steps using LLMs. Built with Python, Supabase, React, and Tailwind. Your roadmap, your way.

🚀 RoadMapr — Your career, charted by AI.

A full-stack platform to write, run, and share code templates, publish dev blogs, and spark threaded discussions. Built with Next.js, React, Prisma, and REST APIs. Features include real-time code execution, JWT auth, search, and content voting.

🧠 Scriptorium — Where GitHub meets Reddit

🤖 Scriptorium — Where GitHub meets Reddit

Engineered a fully functional shell in C with support for job control, process group management, foreground/background execution, I/O redirection, and piping. Implemented built-in commands (fg, bg, jobs, quit) and robust signal handling using SIGINT, SIGTSTP, and SIGCHLD. Designed a clean architecture to manage concurrent processes and system-level communication — recreating key behaviors of modern Unix shells from scratch.