Kaggle Grandmaster Panel at H2O World India
- Generative AI
- H2O Hydrogen Torch
- H2O Driverless AI
Overview
Recorded at H2O World India, this panel brings together four H2O.ai Kaggle Grandmasters, Shivam Bansal, Parul Pandey, Chunming Li, and Pascal Pfeiffer, for a wide ranging conversation on competitive data science. They discuss what it takes to reach Grandmaster status, the personal sacrifices involved, and how skills built on Kaggle, remote collaboration, structuring open ended problems, timeboxing, technical writing, carry over into building H2O's own LLM products. The session closes with audience Q&A on breaking into a new domain, starting out on Kaggle, and the impact of LLMs and AutoML on competitive data science.
Key takeaways
- What separates a Kaggle Grandmaster from a strong competitor: consistency over time, not just one good result
- How Kaggle skills, remote collaboration, structuring open ended problems, and working to a fixed deadline, translate directly into solving customer problems at H2O
- How the panelists approach an entirely new domain: start with the data and the metric before reaching for state of the art models, then expand by reading others' solutions
- Why the panelists see human understanding of the business problem as the most durable skill, even as LLMs and AutoML lower the barrier to entry
- How competing on Kaggle shaped H2O products like Hydrogen Torch, Hydrogen, DocChat, and LLM Studio
- Practical advice for beginners: pick a domain you're genuinely interested in, team up rather than work in isolation, and finish what you start
Start owning your AI