- 🏆 Computer Vision Hackathon Winner 2024 🥇 1st Place at the NorgesGruppen Data × Astar Consulting real-time computer vision hackathon.
- 🚀 Positions:
- I currently serve as the CEO of Cogito NTNU, Norway's largest technical AI student organization.
- Leading and contributing to a portfolio of AI projects across computer vision, reinforcement learning, agentic AI, and decision-support systems.
- Currently: Team Lead & Lead Architect on CancerVision, building a 3D semantic segmentation system for MRI brain tumor analysis.
- Previously: AI Engineer on J.A.R.V.I.S, TrainerAI, and RL Engineer on DeepTactics TrackMania.
- 🏦 Education:
- I'm an M.Sc. student in Materials Technology at NTNU Trondheim, with additional coursework in C++, algorithms, data structures, mathematics, numerical methods, and analytical problem solving.
- 🏐 Outside of Tech:
- I compete as an elite volleyball athlete with NTNUI in the Norwegian Elite Series.
- 💬 Let's Talk About:
I'm passionate about applied AI, computer vision, deep reinforcement learning, agentic systems, MLOps, CI/CD, and building robust AI systems for complex, society-critical problems. I enjoy working at the intersection of technical depth, leadership, and real-world impact.
- Computer Vision systems for medical imaging, object detection, and real-time inference.
- Deep Reinforcement Learning agents for games, autonomous systems, and decision-making under uncertainty.
- Agentic AI applications with tool use, long-term memory, observability, and natural interaction.
- MLOps and AI Infrastructure, including reproducible training, experiment tracking, CI/CD, model registries, and containerized deployment.
Whether you're a seasoned developer, a student, an athlete, a researcher, or someone just starting out in tech, I'm open to collaborating and learning together. Let us build intelligent systems that matter :)
🚀 Cool Projects I've Worked On
CancerVision is a 3D semantic segmentation system for MRI brain tumor analysis using the BraTS 2020 dataset.
The system identifies tumor regions such as tumor core, edema, and enhancing tumor areas across four MRI modalities.
Built with Python, PyTorch, MONAI, DynUNet, and a FastAPI-based web interface for drag-and-drop inference.
The training pipeline includes Weights & Biases experiment tracking, Optuna hyperparameter optimization, Slurm support for HPC training, Docker containerization, UV dependency management, GitHub Actions CI/CD, a JSON-based model registry, and a downstream rule-based tumor classifier.
DeepTactics TrackMania explores deep reinforcement learning agents for Gymnasium environments and the TrackMania racing game.
The training stack combines Rainbow DQN methods such as double Q-learning, prioritized experience replay, dueling networks, noisy nets, and distributional categorical DQN in PyTorch.
My work focused on replay-buffer handling, reward shaping, configuration management, experiment tracking, and rigorous evaluation of agent performance.
J.A.R.V.I.S is an open-source agentic LLM assistant with autonomous task execution, long-term memory, natural voice conversation, and tool use.
I contributed to the LangChain and LangGraph-based agent loop, tool calling, system integrations, environment setup, calendar/task integrations, and observability with Arize Phoenix.
The system uses RAG over vector databases for long-term memory, speech-to-text and text-to-speech for hands-free interaction, Docker Compose orchestration, and a Flutter frontend.
TrainerAI is an LLM-based assistant for personalized workout planning, progress tracking, and time-series analysis of training data.
The project combines LLM prompting and tool use with Flask/Python backend services, a Node.js API layer, MongoDB, and Docker-based deployment.
It gave me early hands-on experience with adaptive, goal-oriented AI agents.
And many more projects, experiments, and collaborations — some public, some private, and some still under development.