Projects

"Duckington Inference Microservice (DIM)" is a Python-based backend system built by Duckington Labs to explore energy-efficient inference for Large Language Models. With the rising environmental impact of AI workloads like ChatGPT, Claude, and DeepSeek, DIM provides a greener alternative by integrating FPGA-based inference alongside traditional GPU inference.

The backend, implemented in Python using boto3 for AWS integration, allows users to run and compare FPGA and GPU inference workloads side by side. DIM offers a user-friendly interface through a Streamlit-based GUI for demonstrations and click-based CLI tool “QuackCLI” for developers and power users. Together, they provide flexible access to real-time performance and energy efficiency metrics.

"MUN Room" is a student-focused rental platform designed to simplify housing searches around Memorial University. Built as a group project, it features a Node.js backend written in TypeScript and uses MongoDB for efficient data storage and retrieval. Students can browse available rooms, view details, contact landlords, and manage their own listings—all in one place. The app streamlines the rental process, making it easier for students to find affordable accommodation quickly and safely.

It is more than just a rental app - it is a community-driven platform for students to connect and find their next home away from home!

"Dear Diary" is your personal space in the digital world for daily journaling. Built with Flutter and Dart, it lets you easily store, edit, and delete your thoughts and memories. Rate entries, upload images, and enjoy a clean list view. Gain insights with stats like entry count and average ratings. Your memories are safe on Firebase Storage.

It's more than a diary—it's a simple, intuitive way to capture and cherish your moments!

Explore cities effortlessly with this Flutter app powered by Dart and Firebase. Discover attractions seamlessly, thanks to Dart's cross-platform prowess, while Firebase ensures real-time updates and a reliable backend.

Your city exploration just got more exciting and user-friendly!

This website, built on the Astro framework, is designed for speed and search engines. It's tailored for recruiters and tech enthusiasts, offering easy access to information about my skills, experience, and projects.

The project prioritizes simplicity by employing reusable components and minimizing reliance on third-party libraries. This approach streamlines development and lowers maintenance efforts.

It also includes sections on Neural Networks with Tensorflow, Deep Computer Vision, Natural Language Processing using Recurrent Neural Networks (RNNs), and Reinforcement Learning using Open AI's Gym module with Frozen Lake v1 for reinforcement learning.

The project implements Cat vs Dog Identifier, Titanic Survival Prediction, Flower Species Classification, Weather Predicting Model, and Play Generator.

Taking inspriation (and frustration) from the Government of NL's ferry services, our group has decided to create a ferry reservation web application. It allows users to view the ferry schedule much easier and enables the additional reservation functionality.

The app allows users to search for their desired crossing, make a booking, check the status of their booking, cancel their booking. The app will also send email confirmation when the user makes a booking.

The game includes an AI opponent with varying levels of difficulty (Dumb vs Smart). The AI employs intelligent strategies to analyze the board and make decisions, providing a challenging experience for players.

The game's design follows SOLID principles, ensuring a clean and modular codebase that is easy to understand, maintain, and extend. Each class has a single responsibility and is highly cohesive, leading to a more robust architecture.

The project included using Linked List, Priority Queues and Heap to create inventory list, customer history. Functionality for automatic inventory reorder was set up, along with sorting inventory by expiry date.

The backend stored the customer and investory information in JSON file. The transactions and orders are executed by reading the JSON file, updating the data structures, and then writing the JSON file back to disk.

The game is a Python application that uses the PyGame library to create a simple game where a bunny defends a castle from an attacking horde of badgers.

The game includes features such as movement, shooting, and collision detection. It also includes a HUD that displays the bunny's health and the number of badgers remaining.

Built using Python software and Arduino kit (button, LED screen, sensors) to create a system to screen for COVID-19 symptoms.

The project uses the teachings of the course content to create a system that combines software and hardware to make a solution to screen for COVID-19 symptoms.

This is the first course in University of Helsinki computer science learning track, and we use industry-grade tooling from day one. As such, the course is about programming in a real integrated development environment (IDE) instead of a browser or an educational IDE.

The course is split up into two individual courses: Java Programming I and Java Programming II. Each course consists of seven parts, each with its own exercise set. Completing a single part takes approximately 5-20 hours.