I'm a founding member at Lens Academy, a course platform that aims to teach tens of thousands of students a month about superintelligence risk and help them find their path to impact. I create and facilitate our courses and build the platform.
I'm a fast-moving, high-autonomy generalist obsessed with automating and optimizing workflows with AI.
AI is probably going to kill us all (~70%). I've dedicated my career to changing those odds and am exploring how AI itself can accelerate alignment.
Course platform teaching superintelligence risk at scale. Founding member: I create and facilitate courses and build the platform.
AI Safety · EducationPodcast exploring cutting-edge research on using AI systems to align other AI systems. Each episode covers a key paper in the field of scalable AI alignment.
AI Safety · SpotifyManage dozens of Claude Code sessions from a single dashboard. See what each is doing, switch between them, and never lose context.
Electron · JavaScriptMeasuring how evaluation awareness affects deceptive behavior in LLMs.
AI Safety · ResearchDrop a PDF, get a podcast on Spotify. Fully automated from research paper to published episode, zero manual steps.
Automation · PythonScrapes thousands of scientific papers, extracts biological interactions, and visualizes them as self-updating causal graphs. Built at Griffin Labs.
Biotech · AILiving and working at the Centre for Enabling EA Learning & Research in Blackpool, UK, together with other people working on the world's biggest problems. Focused work on Lens Academy during the day, communal dinners, circling, and long conversations in the evening.
Lens is a course platform that aims to scalably teach tens of thousands of students a month about superintelligence risk and help them find their path to impact. I work on improving our educational platform and courses on superintelligence risk, AI-safety strategy, compute verification, and theoretical alignment, while also facilitating a bit. The questions I keep coming back to:
Spent May living and thinking with ~40 other alignment-focused people in a castle in Czechia. Deep discussions on agent foundations, theoretical alignment, and AI safety strategy. This is also where I met Luc, the founder of Lens Academy, who recruited me on the spot.
Spent $6,776.74 in Claude API credits this week alone. So far I've built an entire ecosystem of developer tools around Claude Code: a production session manager (Open Cockpit), 3 CLIs, 5 plugins with a custom marketplace, and a fully automated AI podcast pipeline.
Today I finished all courses of my AI bachelor's after just 1.5 years (instead of 3) of being enrolled and with extraordinary grades (summa cum laude).
Griffin Labs is a Biotech/AI startup with the mission to accelerate scientific research towards solving human diseases, specifically focusing on Alzheimer's. As part of my role, I am building an autonomous agent that constructs self-updating causal graphs of biological interactions by analyzing thousands of scientific papers (check it out in the video!).
This month, I attended the "Brain-io BCI" and the "Agentic AI against aging" hackathons. I won both! 😜
In the last three months, I helped to organize a longevity pop-up city in Berlin. We had people from all over the world living together and on our weekend summits we had speakers like Aubrey de Grey and George Church.
After having spent a lot of time thinking about the meaning of life, philosophy, and what to do, I decided on a practical goal: "Making aging and death optional." Since superseded by AI safety. If we don't get that right, longevity doesn't matter.
Today, I uploaded the first episode of my new podcast. I decided to call it Rational Optimization and I want to use it to explore the question: "What is, all things considered, the most rational thing to do?" You can also listen to it on Spotify.
View favorite episodesAfter finishing high school, I decided to travel—so I took my bike and cycled to the northernmost point of Europe. Every evening, I knocked on a stranger's door to ask for a place to sleep, and every night I got one!