_

Hi, I'm Andrei! I'm a software engineer focused on machine learning and deep tech. I enjoy working on challenging problems and turning ideas into practical products, with a particular interest in startups. I recently graduated from a joint double-degree master's program at KTH Royal Institute of Technology and Aalto University in Data Science, and I'm now looking to grow as an engineer and learn more about building startups.

Scroll to explore

Work

Master Thesis Intern

Electronic Arts

January 2026 - June 2026

Stockholm, Sweden

Researched and built an LLM-based agent system to predict CI build failures at scale, achieving a precision of 70% and recall of 84%.
Designed and evaluated multiple architectures such as RAG and GraphRAG for automated CI failure prediction under data scarcity conditions.
Integrated a Neo4j knowledge graph with an LLM agent, enabling structured reasoning over build history.
Conducted rigorous ablation studies across configurations to isolate the contribution of each system component.

Machine Learning Intern

Filed

June 2025 - August 2025

Stockholm, Sweden

Redesigned and built the evaluation pipeline, delivering reliable metrics that empowered the engineering team to optimize prompts and agent system components.
Built a synthetic data generation tool for automated anonymization using LLMs and RAG, helping the startup achieve privacy regulation alignment while accelerating development workflows.

Software Engineer

Garmin

July 2023 - August 2024

Cluj-Napoca, Romania

Developed an end-to-end subscription activation platform using ASP.NET and React with TypeScript, saving Garmin 30% on transaction fees from third-party stores.
Optimized subscription flows by prefetching and caching 50% of the API calls.

Projects

Startup

Ithaca

Legal TechProduct StrategyGo-to-market

Co-founded a legal tech startup with four colleagues, driving strategy, product development, and go-to-market initiatives while validating the market and building the MVP.

AI Product

Caplt

PythonPyTorchFastAPIReact Native

Trained an image captioning model using Python and PyTorch to generate simple descriptions from images, then integrated the model into a social media app built with FastAPI and React Native.

Scientific Computing

n-Body Simulation

C++MPISFMLPhysics

Built a particle simulation system in C++ with a graphical interface using SFML and used MPI to parallelize the physics computations for performance improvements.

Expense Tracking

Money Manager

ASP.NET Web APIReactTypeScriptFull-stack

Designed and developed an expense management application for tracking and analyzing financial transactions.

Web App

Humphrey

JavaScriptMongoDB

Developed a music discovery web application that maps connections between any artist and my favorite artist through song collaborations.