Sessions from Fully Connected 2024 in San Francisco

Want to catch up on the amazing content from 2024 Fully Connected conference? 

You can watch the sessions on demand below. More are being added in the coming days.

2024 Fully Connected conference keynote and session videos


The era of generative AI

Lukas Biewald, CEO and Co-Founder at Weights & Biases

We kicked off Fully Connected 2024 with a keynote from Weights & Biases CEO and Co-Founder Lukas Biewald. He shared his perspective on the generative AI industry: where we’ve come from, where we are today, and where we’re headed.

Tools for building AI applications

Shawn Lewis, CTO and Co-Founder at Weights & Biases

Hear from Shawn Lewis on the new Weights & Biases products and features for foundation model builders, enterprises building and fine-tuning their models, and software developers developing generative AI applications.

How Meta trained Llama 3

Joe Spisak, Product Director, Generative AI at Meta

We were thrilled that Joe Spisak, Product Director of GenAI at Meta, unveiled the latest family of Llama models, Llama 3, at Fully Connected.

Learn all about the training processes and alignment of Llama 3, which now ranks as the top-performing model in the open weights category on the MMLU, GSM-K, HumanEval benchmarks.

Overcoming the complexities of generative AI

Kari Briski, VP of Generative AI Software Product Management at NVIDIA

As models grow in complexity and scope, so does the challenge of training them. And few companies understand those challenges better than NVIDIA.

Join Kari Briski as she walks us through how the most innovative companies in the world are grappling with the challenges of training massive models for real-world use cases. 

The future of trust in LLMs

Richard Socher, Founder & CEO at

Richard Socher, CEO and founder of and AIX Ventures, shares insights from his journey of a decade of research in AI and NLP, from the invention of prompt engineering to founding, the first AI Assistant to integrate an LLM with live web access for accurate, up-to-date answers with citations. Richard discusses how to tackle the biggest challenges facing LLMs, from hallucinations to generic responses. 

Snowflake Copilot: Building the most powerful SQL LLM in the world

Vivek Raghunathan, VP of Engineering at Snowflake

Join Vivek Raghunathan to learn how Snowflake approaches building a powerful large language model for SQL, the challenges they faced along the way, and how they overcame them. 

Generative AI: Scaling Adobe Firefly infrastructure and ML workflows

Ersin Yumer, Sr. Director of Engineering, AI/ML and Data at Adobe

At Adobe, hundreds of researchers and engineers work on large-scale generative AI models from initial research and prototyping to production and serving models at inference time for many applications to consume, including Adobe flagship products such as Photoshop.

In this talk, Ersin introduces how Adobe revamped and scaled their ML infrastructure and workflows to optimize for speed in research to production.

Getting started with large-scale training and LLMOps in Azure AI

Manash Goswami, Principal Group Program Manager at Microsoft

Azure AI is at the foundation of generative AI innovations you see today: OpenAI’s ChatGPT and Microsoft Copilots are all built on top of Azure AI platforms and tooling. 

In this session, Manash covers a wide range of topics and best practices from large-scale training to build your LLMs and leverage pre-built LLMs from our model catalog. You’ll also learn how to get from prototype to production with LLMOps and new developer tools for iterative debugging, evaluation, deployment, and monitoring.

Deploying one of the first NVIDIA GH200 Grace Hopper Superchip Clusters in Lambda Cloud

David Hall, VP of NVIDIA Solutions, Lambda

Lambda introduced one of the first NVIDIA GH200 GPU clusters in its cloud, featuring NVIDIA’s new ARM-based Grace CPU for enhanced efficiency and coherent NVIDIA NVLink-C2C interconnect that provides 900 GB/s of bandwidth between the Grace CPU and Hopper GPU. The presentation will discuss Lambda’s GH200 cluster design optimized for ML training and our findings on training performance.

Training recipes and scaling strategies for high-quality GenAI models

Natalia Vassilieva, VP and Field CTO, ML at Cerebras Systems

Join Natalia as she shares how Cerebras trains large models and what they learned along the way. She shares her experience and insights from training various LLMs and multi-modal models, techniques for compute-efficient training of dense models, amd the benefits of sparse training and inference on Cerebras hardware.

AI 2024: The journey from here

Sri Viswanath, General Partner & Managing Director at Coatue

In this closing keynote, Sri gives an investor’s point of view about the state of AI today and where he expects our space to be in the next year. He looks at broad investment trends, the transition from research to deployment to business value, and forecasts the next big thing in AI. 

NVIDIA and Weights & Biases fireside chat

Lukas Biewald, CEO and Co-Founder at Weights & Biases + Manuvir Das, VP of Enterprise Computing at NVIDIA

Join Manavir and Lukas for a candid conversation about the direction of machine learning as a space, why NVIDIA is investing so much in NIM, the partnership between NVIDIA and Weights & Biases, and what both men expect to see in the coming year in AI. 

Breakout sessions

The subtlety of stepped change: how Mercari is using Gen AI

Teo Zosa, Senior Machine Learning Engineer at Mercari

Mercari revolutionized the e-commerce industry in Japan almost immediately after its founding to become Japan’s largest C2C online marketplace. In service of our commitment to delivering the highest quality product to our users, Mercari is using AI in subtle but significant ways to create outsized customer value and redefine Japan’s e-commerce landscape.

Continuous Deployment with W&B Automations

Ali Demirci, Senior Machine Learning Engineer at Rad AI

Rad AI generates thousands of radiology reports daily through its online inference systems. We will talk about how we utilize Weights & Biases Model Registry and Automations to build and deploy machine learning models (including LLMs).

Building a High-Performance Computing Cloud Platform

Rahul Talari, Sr. Machine Learning Platform Engineer + Harsh Banwait, Director, Product at CoreWeave

Discover how architectural design, GPU-acceleration, and scalability play crucial roles in supporting cutting-edge AI projects. Additionally, we will explore how a collaboration with Weights & Biases can augment experiment tracking and model optimization, offering a holistic approach to AI development.

Human x Machine: From Models to Products

Seth Levine, Lead Machine Learning Scientist at Loris

This talk goes behind the scenes on the models, systems, and special sauce that power our user-centric offerings. It’s about finding the right human-AI balance to build products customers love.

Robotics AI for Industrial Applications

Kyle Coelho, Research Engineer + Brian Zhu, Research Engineer at Siemens

In this session, Siemens’ research engineers share their expertise on applying AI to robotics for industrial applications. From CV to RL, discover how these technologies are enabling automation in manufacturing and logistics, addressing both technical challenges and economic motivations. 

Samba-1, Enterprise Grade Open Source AI

Anand Sampat, Senior Director, Machine Learning at SambaNova Systems

Anand discusses how SambaNova’s cutting-edge hardware and platform integrate to deliver robust, secure, and cost-efficient AI solutions. Learn how Samba-1 enables enterprises to harness the power of both large and small models, providing flexibility and precision in various applications.

Understanding LLM performance in Snowflake using Weights & Biases and Snowpark Container Services

Vino Duraisamy, Developer Advocate, Snowflake at NVIDIA

Join Vino Duraisamy at the Fully Connected conference for an in-depth look at LLM performance in Snowflake, utilizing Weights & Biases and Snowpark Container Services. Vino shares valuable insights on fine-tuning and evaluating models, backed by a live demo and practical examples. This session is ideal for anyone interested in harnessing Snowflake’s AI and generative AI offerings to enhance their data science and machine learning projects.

Thank you to our sponsors!

Thank you to our sponsors!