Machines and Molecules

A decade ago, Martin Rahmel co-founded a startup with the idea of dissolving catalysts in water for reuse. Today, he leads the Chemical Invention Factory (CIF) in Berlin, a hub for green chemistry startups. CIF's story began simply with providing a brick-and-mortar lab to decrease cost for chemistry startups, now blossoming into a ecosystem of innovation with 10Mio Euros of funding. Martin shares the trials of translating science into business, recalling a year without sales before breaking through with their first client, Lonza. His insights on the mindset needed for startup success are invaluable, emphasizing perseverance and confidence. Beyond CIF, Martin's passion extends to restoring river systems, showcasing his dedication to environmental change. Join us as we explore how self-reflection and an objective perspective are key to driving scientific and sustainability advancements.

This episode features Max Welling, a visionary at the intersection of science and machine learning. Starting under Nobel laureate Gerard 't Hooft, Max shifted from the enthralling world of physics to machine learning, inspired by the methods of his early studies. His journey through prestigious labs, including Turing Award winner Geoff Hinton's, illustrates his belief in intuition-driven research. For a long time now, Max has been pioneering ML methods using a broad toolkit of approaches. These days, he is blazing a trail in applications of ML in climate change and scientific advancements. His resilience, shaped by youthful dreams of aviation and overcoming his disappointment in not becoming a fighter jet pilot, fuels his relentless pursuit of knowledge and innovation. 

In this enlightening episode, we delve deep with Finnish machine learning grandmaster Aapo Hyvärinen, as he discusses his latest book, “Painful Intelligence”. Hyvärinen, a trailblazer in unsupervised learning, sheds light on the profound connections between human suffering, the pursuit of goals, and the mechanics of learning. His discussion moves from super technical realms of non-gaussian latent variable models to the philosophical, exploring how the connection between machine learning and mindfulness meditation. As we venture into the future of AI, Hyvärinen raises thought-provoking questions about possibilities for the emergence of a survival instinct of AI agents.

In this episode, we explore the challenges faced by scientists like Patrick Torbey, the founder of Neoplants, as they transition from academia to entrepreneurship. Patrick shares his struggles in finding a co-founder with business acumen, emphasizing the rare intersection of academic and business worlds. He discusses how building a startup, like his venture that engineers plants to purify air rapidly, means strong improvements in terms of speed and resources compared to academia. Patrick underscores the importance of scalability and reproducibility in the journey from lab to market, and how the consumer can be a very harsh critic

In the latest episode, Alan Aspuru-Guzik introduces us to the transformative potential of self-driving labs, where complex chemical equipment merges with machine learning. However, it's not about chasing the perfect algorithm – it's about making things work. Old workhorses like Bayesian optimization he augments with a wide range of innovative twists. Motivated by David King's chilling take on climate change, Aspuru-Guzik urges us to venture into uncharted territory, balancing imagination with impact. Because, in the end, it's about doing what can truly add value.

In this captivating episode, delve into the intriguing world of enzymes, nature's microscopic machines, as Uwe Bornscheuer recounts his early fascination with their intricate design and function. Traverse through time as we explore both historical and contemporary methods of enzyme engineering and discover the parallels between enzyme optimization and techniques used in machine learning and operations research. Uwe's driving forces are a potent mix of impact on the world and the joy of nurturing people. And for a lighter note, Uwe reveals his 'M-Word' – his (not so secret) escape from the lab. Don't miss it!

In this enlightening episode, Kristof Szalay unveils the journey of Turbine - from a mere idea to a pharmaceutical game-changer. By simulating how proteins send messages in cells, Turbine is on the cusp of slashing drug discovery timeframes and ushering in groundbreaking drugs. Kristof highlights how Turbine transitioned from academia's incremental approach to the realm of engineering moonshots. The discussion delves into the sweet spot between innovation management and unfettered freedom in realizing one's vision. Dive in to learn why a "just do it" approach carries you far in audacious projects.

This episode is a deep dive with Tina Klüwer, a beacon in Germany's AI landscape. From trying to understand human thought by replicating it in machines to founding an AI startup in 2014, Tina's overflowing enthusiasm has always been an asset. We discuss the nuances between academia and startup culture and tackle pressing topics like societal biases mirrored in AI. We touch upon her role in explaining to Germany's policymakers what AI can and can't do, as well as the question of whether machines have developed consciousness.

In this episode, Ingmar chats with François-Xavier Briol (FX for short), a lecturer at University College London. FX specializes in probabilistic numerics and its applications in the natural sciences. In the episode, we nerd out about model misspecification, which addresses the challenges posed by the idea that "all models are wrong, but some are useful." We also discuss personal growth and the importance of collaboration in FX's academic career.

In this first episode of Machines and Molecules, Ingmar talks with Dr. Mikio Braun talks about machine learning before it was a hype, Math and persevering through hard phases in your work (read: PhD thesis). We discuss what you learn when you move from academia to industry and the probability that humanity will be terminated by AI.