In addition to the tech talks, there will be plenty of opportunities to network with AI developers, live demos by AI startups, panel discussion, and career opportunities.
- 5:30pm-6:20pm: Checkin, food/drink and networking
- 6:20pm-6:30pm: Welcome, Community update
- 6:30pm-8:30pm: Tech talks, and panel
- 8:30pm-9:00pm: Q&A and Open discussion
Tech Talk 1: Practical Lessons from Ads Ranking
Speaker: Aayush Mudgal @Pinterest
Abstract: Join us for a talk on the journey of scaling ads ranking at Pinterest using innovative machine learning algorithms and innovation in ML platform. This presentation will showcase the transition from traditional logistic regressions to deep learning-based transformer models, incorporating sequential signals, multi-task learning, transfer learning. Throughout the process, we encountered various challenges and gained valuable lessons. Discover the hurdles we overcame and the insights we gained in this talk, as we share the transformation of ads ranking at Pinterest and the lessons learned along the way
Tech Talk 2: Snakes in Metaflow
Speaker: Romain Cledat @Netflix
Abstract: Metaflow helps data-scientists be more productive and focus more on data-science and less on infrastructure. At Netflix, we want to enable data-scientists to independently innovate, in part by selecting the libraries and packages they want for their problem, but also to share results with one another and be able to reproduce others experiments. This, it turns out, is surprisingly complicated. In this talk, we will first introduce Metaflow before diving into the viper pit of providing a reproducible environment in the presence of external dependencies. We will see how Metaflow solves this by relying on additional snakes to rule them all.
Lightning Talk 1: LLMs Hallucinate and RAG
Speaker: Ofer Mendelevitch @Vectara
Abstract: In this talk I will discuss why Hallucination occurs, and some of the ways to address it, including retrieval augmented generation (or RAG), and discuss the pros and cons of each approach.
Lightning Talk 2: Oracles and Worker Bees: Why Small Models are the Future
Speaker: Emmanuel Turlay @Airtrain
Abstract: In this talk we look at evidence that small models are where developers, startups, and enterprises should focus their resources in order to accelerate safe deployments of AI