Francesco
Lässig

Building sovereign AI infrastructure for Swiss companies: open-source models, on-premise & Swiss cloud.

• AI Learning Coach for Entlastungsdienst Bern: multi-agent system (5 specialised agents) for scenario-based competency assessment, personalised learning paths, interactive training, and a confidential reflection space for care workers. Delivered in 14 weeks (Nov 2025 – Feb 2026).
• Kontexus: co-developed and rolled out an on-premise / Swiss-cloud platform based on open-source LLMs, deployed across multiple client industries.
• Semi-finalist, GenAI Awards 2026.

Built LLM applications for clients in tech, finance, and energy.

• Text-to-Data Chatbot (major US tech company): multi-agent conversational analytics system for querying business and sales data in natural language — identified the correct database, executed queries, generated grounded answers and dynamic charts.
• Data Quality Monitoring (Swiss energy company): end-to-end anomaly detection pipeline for operational data; led stakeholder requirements workshops and a comparative evaluation of candidate solutions.
• LLM Due Diligence Assistant (Swiss asset manager): semantic document retrieval system for due diligence workflows with full source traceability — every answer linked to its origin document.

• Developed a significant part of Darts, an open source library for time series forecasting, including statistical and deep learning-based forecasting tools. Presented Darts at the EuroPython 2021 conference and the PyData Global 2021 conference. During the time I worked on Darts, its GitHub page went from 0 to over 3.3k stars.
• Built a ML-based predictive maintenance tool for a Swiss hydro power plant, all the way from exploratory data analysis and model development to backtesting and deployment.
• Developed a demand forecasting solution for a Swiss manufacturer of laboratory and industry equipment which improved their existing forecasts by 10% - 50% (depending on the metric).
• Co-hosted multiple technical public webinars revolving around topics in data science and machine learning.

• Developed a novel, bio-inspired continual learning algorithm called sparse-recurrent DFC as part of my master thesis, which received the maximum grade.
• Showcased poster about my master thesis at the AI+X Summit 2022. Presented my work at an IROS 2022 workshop on continual learning.
• Founded Qualiaheads, a student club and reading group on consciousness science. Conducted interviews with researchers such as Anil Seth and Pedro Mediano.
• Finished degree with a weighted GPA of 5.8 out of 6.

• Took courses at the computer science department and the Wharton business school.
• Received honorable mention for Facebook-sponsored award in a project-based coding competition as part of the NETS 212 course (among top 4 of 54 teams).
• Finished the semester with a GPA of 3.75 out of 4.

• Worked as a student assistant teaching calculus.
• Received a scholarship for a selective exchange program to the University of Pennsylvania.
• Completed degree with a GPA of 5.36 out of 6.