MSc Computer Science
Thesis: Agentic LLM Systems for Automated Scientific Discovery (at ETH Zurich)
Member of the advisory board of the CS study association.
Lorenzo Steno
MSc Computer Science · University of Twente
Incoming Master Thesis Student at ETH Agentic Systems Lab
About
I am a Master's student in Computer Science at the University of Twente, starting my thesis at ETH Zurich's Agentic Systems Lab in February 2026.
Previously, I worked as a Solutions Architect Intern at AWS and as a Research Assistant at the AI & IoT Lab at The University of Twente. I focus on post-training methods for large language models, including fine-tuning, reinforcement learning, and agentic systems, with an emphasis on safety and trustworthiness.
Research Interests
- Multi-agent LLM systems and autonomous scientific discovery
- Reliability and trustworthiness of large language models
- Reinforcement learning for complex decision-making
Education
BSc Computer Engineering
Experience
Master Thesis Student
Designing autonomous agentic systems using LLMs for systematic literature reviews. Focus on multi-agent workflows for large-scale corpus synthesis and reliability in automated scientific discovery.
Research Assistant
Researched applications of generative AI in digital twins and distributed systems. Designed and deployed cloud-based experiments on AWS.
Solutions Architect Intern
Implemented GenAI capabilities into internal tooling, automating summarization and tagging (~40% reduction in review time). Developed serverless workflows using Lambda, S3, DynamoDB, and Bedrock.
Teaching Assistant
Guided 30+ students through GPS-to-MQTT IoT pipelines. Managed AWS IoT Core, EC2, and IAM policies for the AI and IoT Lab.
Selected Projects
Green AI: Energy Efficiency of PEFT Methods
Co-authored a study evaluating energy-to-performance trade-offs of parameter-efficient fine-tuning for code generation using Qwen 3. Developed energy measurement pipelines with NVIDIA's NVML.
Energy-Efficient ML PEFT NVML
Clash Royale Reinforcement Learning Agent
Developing a model-based RL agent (DreamerV3) to master complex real-time strategy gameplay. Engineering a visual perception pipeline using YOLO for real-time game state extraction.
Reinforcement Learning Computer Vision DreamerV3
Agentic AI Assistant for Compliance Analysis
Built an AI prototype for ING bank to assist with compliance analysis. Implemented interoperability and traceability using the Model Context Protocol (MCP).
LLM Agents MCP Compliance
Certifications
- AWS Certified Solutions Architect Associate
- AWS Certified AI Practitioner
- BlueDot Impact – Technical AI Safety
- Santa Fe Institute – Introduction to Complexity