The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) Architectural and Organizational Patterns in Machine Learning with Nishan Subedi - #462
Mar 8, 2021
Nishan Subedi, VP of Algorithms at Overstock.com, shares his journey from physics to leading machine learning initiatives. He delves into how Overstock utilizes ML for search and marketing, highlighting the importance of architectural patterns in ML systems. Nishan discusses the innovative concept of 'squads' in organizational structures and how flexibility and collaboration enhance team effectiveness. He also examines the challenges of moving ML from the lab to production and the future of integrated architectural patterns in the industry.
AI Snips
Chapters
Transcript
Episode notes
Nishan's Career Journey
- Nishan Subedi's career started in systems engineering and DevOps before transitioning to search at Etsy.
- He was drawn to search by the challenge of improving user experience through relevance.
VP of Algorithms
- The title "VP of Algorithms" reflects a focus on building algorithmic products, not just applying ML/AI.
- It emphasizes the importance of starting with simple solutions before incorporating more complex methods like ML/AI.
Architectural Patterns in ML
- Architectural patterns in machine learning, like those in software engineering, involve understanding the forces shaping the problem.
- Applying ML models in production often leads to key learnings and adjustments due to real-world constraints.