Hi, I'm Thomas 👋
Student and developer with a passion for AI, web development, and entrepreneurship.
TN

About me

I love building solutions that solve real problems and continuously learning new technologies. Currently studying Computer Science at NTNU and working part-time as a security guard. In my spare time, I lead an AI project in collaboration with Infor, where I've worked on recommendation systems, price recommendations, and experimented with AI agents. I'm always looking for new challenges and exciting opportunities!

Skills

HTML
CSS
JavaScript
TypeScript
React
Next.js
Node.js
Python
Framer Motion
Java
Figma
JavaFX
Git
Tailwind
My projects

Check out my latest projects

I've worked on various projects, from simple websites to complex web applications. Here are some of my favorites.

Futleie

A rental website developed in the Software Development course. Focus on user experience, secure login, and simple administration of rental objects. The work provided good insight into planning, coding, and testing a complete web application.

Next.js
Tailwind CSS
Supabase
Fullstack web development
Software development

Cogito x Infor

Developing recommendation systems and price recommendations for customers based on previous purchase data. Experimenting with AI agents to optimize internal business processes. Responsible for both technical implementation and team leadership.

Machine Learning
Python
ERP Systems
Project Management
AI Agents
Recommendation Systems

News Recommendation System

Developed a content-based recommendation system for news articles that uses sentence transformers to find similar articles based on user preferences. Responsible for the entire pipeline from data collection to implementation of the recommendation algorithm.

Python
Sentence Transformers
Machine Learning
NLP

Kollapp

App for students in shared housing that simplifies daily life by making it easy to share tasks like garbage and cleaning, handle settlements when splitting food costs, and communicate via integrated group chat. Automates many of the challenges students in shared housing face.

Java
JavaFX
Hackathons

I love building things

During my studies, I have participated in 5 hackathons. People from all over the country gather to build incredible things in 2-3 days. It's inspiring to see the endless possibilities created by committed and passionate people.

  • R

    Recover Norge Hackathon

    Trondheim

    We won the Recover Norge Hackathon by developing a machine learning solution to predict missing actions in repair calculations. Recover Norge helps people when their homes are damaged, but their team often spends hours manually identifying missing actions. We built an ensemble model combining XGBoost, CatBoost, LightGBM, and a custom neural network to automate this process. For the final task, we created three live production demos in just a few hours based on real problems Recover faces daily, demonstrating practical tools they could immediately use to speed up their workflow and focus on fixing homes instead of fixing calculations.
  • N

    Norwegian AI Championship 2025

    Norway

    We won 2nd place in the Norwegian AI Championship 2025 with Cogito, a week-long competition featuring three challenging tasks: NLP with retrieval-augmented generation for medical documents, reinforcement learning with a self-driving car, and computer vision with tumor segmentation.
  • A

    A* Hackathon 2025 w/ Norges Gruppen & Cogito

    Trondheim

    During a 16-hour hackathon organized by A* Consulting in collaboration with Norges Gruppen and Cogito, we developed a real-time AI system (YOLOv11) for automatic product classification, receipt generation, and theft detection in self-checkout systems. Our live demo using mobile camera, web app, and a robust computer vision model secured us a solid third place.
  • B

    BRAIN Hackathon 2025

    Trondheim

    Organized by DNV with a focus on developing an object detection model for autonomous vessels. The task was to create a machine learning model for obstacle detection – a critical step in developing autonomous collision avoidance. The competition was conducted on Kaggle, with 24 hours to achieve the highest possible accuracy, while emphasizing model interpretability and risk assessment. We ended up in second place and were awarded best presentation and most creative approach. We used computer vision models like YOLO 11.
  • S

    Start Code 2024

    Trondheim

    Participated in a 48-hour hackathon organized by Start Code, where we solved a case from one of their main partners, emno. We worked on automatically connecting a Raspberry Pi to a website. It was an intense and educational experience that provided insight into both hardware integration and web development under time pressure.
Contact

Get in touch

Want to chat? Feel free to reach out via LinkedIn or send me an email at thomanso@stud.ntnu.no.