From Data Lakehouse to Data Flywheel
Modern AI systems don’t just learn once. They learn continuously. Yet building data infrastructure that supports this safely and at scale remains one of the hardest challenges for organizations today.
This session explores how leading AI teams design data pipelines and feedback loops that measure performance, capture feedback, and build scalable data flywheels for responsible model improvement.
Drawing from her experience at OpenAI, Google, and Twitter, Swetha Sekhar will share how to build the foundations of continuous learning, turning every model interaction into a measurable, safe signal for improvement.
Key Topics
✦ Continuous learning in production environments
✦ Logging principles for safe and auditable AI systems
✦ Synthetic data and ablation studies for model robustness
✦ Avoiding drift and ungoverned retraining
✦ Balancing innovation, privacy, and scale
Why Attend
A concise blueprint for building responsible AI systems that learn and improve over time. You’ll walk away with actionable frameworks to design scalable, safe, and continuously evolving data architectures.
Speaker
Swetha Sekhar
Data Engineer, OpenAI
Swetha builds data infrastructure for classifier development, model evaluation, and safety systems. Before OpenAI, she worked at Google on YouTube Ads Planning and Measurement, with deep experience in analytics and data engineering.
Host
Boaz Descalo – Founder, Node8.ai
Advises mid-market and PE-backed companies on AI strategy and implementation across internal processes and product development. Creator of the Build AI community.
About our Communities
Build AI Community
A network of builders, founders, and technical leaders driving hands-on innovation in AI. Build AI hosts workshops, webinars, hack nights, and collaborative projects that give members the tools, code, and network to take AI from idea to impact.