Data science intern
Cantellat solutions
- Developed and contributed to multiple Model Context Protocol (MCP) servers and an MCP gateway, creating a
- standardized integration layer for AI agents to discover, route, and execute tools across enterprise services.
- Developed AWS CodeCommit MCP integrations and conducted in-depth research on federated identity, JWT
- authentication, Microsoft Entra ID, and AWS IAM Roles Anywhere, translating findings into implementation guidance
- and supporting authentication design decisions.
- Implemented a hybrid RAG pipeline combining vector and keyword retrieval to balance semantic relevance with
- exact-term matching for LLM-powered workflows.
- Researched vector-indexing and retrieval strategies for large codebases, evaluating approaches for scalable semantic
- code search and efficient context retrieval for AI agents.
- Currently contributing to an agentic UI-generation system, developing workflows that translate natural-language
- requirements into functional interfaces through iterative AI-driven generation and refinement.
AI developer intern
Gaddr AB
- Designed and implemented a semantic search microservice in Python, integrating LLM-based embeddings into an existing Next.js backend via FastAPI to replace keyword-based search with context-aware retrieval
- Built and curated a custom sample dataset to train and validate the semantic search pipeline, ensuring relevant and accurate results across the platform's search use cases
- Containerised the Python microservice using Docker and connected it to a PostgreSQL database, enabling reliable deployment within the existing full-stack infrastructure
- Delivered a working end-to-end semantic search feature across the full stack (Python, FastAPI, PostgreSQL, Docker, Next.js), taking ownership from architecture design through to integration and testing
AI developer intern
Realdini studios
- AI Integration: Develop and implement AI-driven systems for NPC behavior, storytelling, and procedural content generation.
- Research & Experimentation: Stay ahead of the latest AI trends and test new models to enhance gameplay experiences.
- Unreal Engine Development: Work with AI systems like Behavior Trees, Blackboards, and procedural AI tools.
- Generative AI Applications: Use tools like Stable Diffusion, LLMs, and OpenAI APIs for asset creation.
- Optimization & Performance: Ensure AI implementations balance innovation with performance efficiency.
master thesis student
axis communications ab
- Developed a deep learning prototype to automate grouping network audio devices using sound similarity analysis.
- Validated and tested model performance with a custom dataset, simulating real-world room setups.
- Applied Python, Machine Learning (ML), Deep Learning (DL), Linux commands, and other technologies
Enhanced Text-to-Speech with Kokoro-TTS
A user-friendly Streamlit application for converting text into high-quality speech using the hexgrad/Kokoro-TTS model. Select from a variety of voices to customize the audio output. Features: Text-to-Speech Conversion: 🗣️ Converts user-provided text into speech. Multiple Voices: 🎤 Offers a selection of American and British male and female voices. Audio Playback: 🎶 Includes an audio player with advanced controls to listen to the generated speech. Download Audio: 💾 Allows users to download the generated audio in WAV format. Responsive Interface: 📱 Built with Streamlit for a seamless user experience on different devices.
Decision Support System for Cryptocurrency Predictions
Designed and implemented a decision support system to predict cryptocurrency market outcomes over short time horizons. Trained and evaluated multiple ML models including decision trees, logistic regression, and random forests to identify the best-performing approach for price prediction. Deployed each model to a production environment using Streamlit, creating a user-friendly web interface for real-time predictions and model comparison.
master of technology
Blekinge institute of technology
Attendee · Oct 24, 2026