I’m an AI engineer based in Jönköping, Sweden, with an M.Sc. in Artificial Intelligence from Jönköping University.
I have hands-on experience developing applied AI solutions across computer vision, NLP, information extraction, graph neural networks, optimisation, and full-stack AI applications.
I’m currently an Artificial Intelligence Intern at Elajo Elektriska AB, where I am extending an AI-assisted warranty claim prototype towards production readiness through backend feature development, .NET/C#, React, CI/CD workflows, and version control.
- Machine Learning: scikit-learn, XGBoost, Random Forest
- Deep Learning: PyTorch, TensorFlow, Keras, CNNs, Vision Transformers
- Computer Vision: Image classification, embedding retrieval, CLIP, DINOv2
- NLP and LLMs: Information extraction, structured data generation, Gemini API
- APIs and Applications: FastAPI, Flask, Streamlit, React, REST APIs
- Data: SQL Server, Pandas, NumPy, FAISS
- Tools: Git, GitHub Actions, Docker, Jupyter Notebook, VS Code
Dockerized NLP pipeline for extracting alloy compositions and material properties from aerospace research papers.
LLM-powered Streamlit application that uses Gemini to extract structured material-property information from scientific PDFs.
TensorFlow experiments using custom multi-head Graph Attention Networks to predict future positions from graph-structured trajectory data.
Machine-learning proof of concept developed with Husgruppen using feature engineering, data augmentation, model comparison, and cross-validation.
Computer-vision prototype developed with Trackmate to classify EUR and non-EUR pallet images using pretrained CNN architectures.
Developed in collaboration with Elajo and Jönköping University.
The prototype combines barcode recognition, DINOv2 and CLIP embedding retrieval, FAISS similarity search, FastAPI, React, and SQL Server to support product identification and decision-making in industrial warranty workflows.