Projects

Working towards a tractable, coherent model of plausible AI futures along their breadth and the genuine Knightian uncertainty surrounding them and how that uncertainty shapes expectations and thus behavior.
Through that channel, the framework aims to illuminate the decision problems faced by investors and policymakers alike: how the perceived range of trajectories, risk aversion, and ambiguity aversion jointly determine prudent allocations of savings, capital, attention, and public resources under deep uncertainty about transformative AI.
Intended outputs: a research blog post(s), potentially a technical working paper.

Open source scaffold for human-AI collaboration (specifically LLMs).
Part project management and documentation template for collaborative knowledge work, part user guide for best practices in human-AI interaction as well as self directed experimentation and learning
Repository

Leveraged Shiny for Python to create an interactive web application that tracks and visualizes the data from the Ukraine Support Tracker of the Kiel Institute for the World Economy (IfW)
Including no-code ETL (Excel to duckdb) pipeline, interactive visualizations, and deployment on shinyapps.io.
Repository,Hosting

Contributed to the development of a scientific Python package for deep reinforcement learning in electricity trading, including formal modeling, algorithm and ETL pipeline design, MongoDB deployment, and HPC cluster orchestration using Ray.