The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) Supporting Rapid Model Development at Two Sigma with Matt Adereth & Scott Clark - TWIML Talk #273
Jun 11, 2019
In this engaging discussion, Matt Adereth, Managing Director at Two Sigma, and Scott Clark, CEO of SigOpt, dive into the complexities of model development in finance. They explore Two Sigma's innovative end-to-end modeling platform and the challenges of production. The conversation highlights how optimization and experimentation tools like SigOpt can transform machine learning processes. The speakers also discuss the motivation behind companies adopting automation, the significance of domain expertise, and the ongoing evolution of programming languages in data science.
AI Snips
Chapters
Transcript
Episode notes
Serving Diverse Modelers
- Two Sigma serves a diverse range of modelers with varying technical skills and domain expertise.
- Their platform aims to support this diversity without being prescriptive about tools.
Time Series Focus
- Two Sigma treats all data as time series, emphasizing point-in-time simulations and backtesting.
- They prioritize time series analysis in both preliminary data analysis and model evaluation.
Platform Goals
- Focus on getting better answers faster when building modeling platforms.
- Prioritize both improved techniques and usability to accelerate the process.