ODSC AI East 2026 Conference | Open Data Science Conference

__________________________________

ODSC AI East will feature 250+ speakers, instructors, and practitioners, 2,500 of your fellow data and AI practitioners, and hands-on instruction for world-class experts in:

With five passes to choose from, you can personalize your experience and get the most out of ODSC AI East:

Silver Pass - 2 days (Wednesday and Thursday) - access to ODSC Keynotes, Talks, and Tutorials, AI Expo & Demo Hall, The Gen Ai X Summit, Networking Reception, Meet the Experts, Dinner with Data Scientists, and other events.

Gold Pass - 3 days (Tuesday - Thursday) - everything included in Silver Pass, plus one Extra day, full of Talks, Tutorials, and Hands-on Workshop Sessions, AI Expo & Demo Hall on Tuesday, and other events.

Platinum Pass - 3 days (Tuesday - Thursday) - everything included in Gold Pass, with the additions of Deep Dive Training Sessions, held by industry leaders, and other events.

AI Mini-Bootcamp Pass - 4 days (Monday (virtual only), Tuesday - Thursday) - everything included in the Platinum Pass plus access to Virtual Bootcamp Fundamentals & Pre-Conference Training and a full year of premium subscription to the Ai+ Training Platform.

VIP Pass - 4 days (Monday (virtual only), Tuesday - Thursday) - everything included with the Mini-Bootcamp Pass, with the additions of daily access to the Speaker/VIP Networking Lunch, Speaker/VIP Lounge, and VIP Fast Lane Registration & Gift Package.

IMPORTANT NOTE: All ODSC AI East Passes are strictly non-refundable if the pass holder has attended Agentic AI Summit or Any Pre-conference Bootcamp Sessions.

You’ll also be able to build meaningful connections with speakers and peers in a relaxed setting during our networking events like the ODSC Networking Reception, Dinner with Data Scientists, Meet the Speaker session, Book Signing sessions, and more.

So, are you excited about ODSC AI East in Boston? Register now and experience the leading technical AI conference for yourself!

What can you expect from an ODSC AI conference? Check out this highlights video from ODSC AI East 2025. ODSC, where the data community connects.

• Ticket prices and availability are subject to increase or decrease, at the discretion of ODSC AI, before and/or after you have made your purchase, and do not entitle the purchaser to a partial refund or credit.
• Given the nature of live events, changes to specific sessions and/or presenters may be unavoidable and do not entitle the purchaser to a partial (or full) refund or credit. ODSC AI reserves the right to expel and/or void the Ticket of any attendee or participant who violates the .

Info: This session cuts through the noise to provide a hands-on, practical guide for product, business, and technical practitioners tasked with validating LLM-powered systems. We will explore a comprehensive evaluation toolkit that balances cost, speed, and quality. We'll cover the pros and cons of automated metrics (e.g., ROUGE, BLEU), the strategic use of LLM-as-judge frameworks for nuanced qualitative assessment, and best practices for designing efficient and scalable human-in-the-loop evaluation pipelines.

Info: In this 30 minute session, AI veteran Cameron Turner will guide the audience through 5 rules all AI practitioners and decision makers should adhere to when building scaled industrial AI systems. Spanning technical and organizational topics, he will share wins and pitfalls Fortune500/Global1k companies have made in early generative AI adoption and offer practical steps to reduce risk and ensure value creation. With an eye toward sustainability and ethics, Turner will describe how to include governance and human oversight into AI release processes to improve AI performance and ROI for any enterprise. Attendees will leave with a better understanding of the current trends in enterprise AI in 2025, and a practical toolset for building/deploying AI in their environments in 2026.

Info: Session will include design blueprints/demo, and lessons learned from real production rollouts—enough to start building your own ecosystem, but with plenty still under wraps for the hallway track. Join us to glimpse the agentic future and leave with the roadmap to reach it.

Info: 2025 is the year of agent: they go beyond basic LLM question-and-answer patterns into an entirely new programming paradigm in which software is given an objective and a set of tools and figures out on its own how to achieve that goal. Agents can not only produce dramatically better, more accurate results in response to queries, but they can be integrated directly into software to accept structured input and take action "under the hood", without needing a chat interface at all. This talk describes practical patterns for building real-world agents including chaining, parallelization and self-reflection, and how to build them in LlamaIndex, with pointers to going deeper to your own "deep research" style agent.

Info: This talk explores known cases, the fraud tradecraft employed, open data sources, and how technology gets leveraged. There are multiple areas were multimodal agentic workflows (e.g., based on BAML) play important roles, both for handling unstructured data sources and for actions taken based on inference. Moreover, we'll look at where data professionals are very much needed, where you can get involved.

Info: This session is to teach you the best practices of applying llama to your domain. We will learn the best approach to using llama out of box on your domain to get the best performance.
The session is split into two parts each are 1 hour long and 25% theory, 75% coding time

We will cover automatic prompt optimization using our (entirely) open source repository that helps you minimize engineering time and allows you to focus on prototyping instead of prompt engineering.
After maximizing out of box performance, we will focus on the best practices of fine-tuning from cutting edge research and learn how to use synthetic data for your domain.

Info: Join this hands-on workshop to build production-ready agents using BeeAI's framework, then deploy them with out-of-the-box web GUIs through the BeeAI Platform. By session end you'll have built functional agents with built-in observability, memory management, and access to tools, all interactive through an easy-to-configure GUI. Perfect for developers who want powerful AI agents applications without building from scratch.

Info: Within five years, AI agents will handle entire feature implementations and production issues autonomously, fundamentally changing what it means to be a software engineer. The traditional individual contributor role will disappear as engineers upskill to delegate, create scalable architectures, and ensure systems operate autonomously. Success will require staying curious and adaptable - continuously learning new tools as models improve. This talk covers two critical areas: reskilling strategies for technical leaders and engineering managers, plus the technical infrastructure needed for human-AI collaboration, including making codebases "agent-ready". Drawing from experience at xAI and building fairgo.ai, you'll gain insights on how to lead in the age of AI agents.

Info: Description coming soon!

Info: Powered by 100 million miles on public roads and billions more in simulation, the Waymo Driver is the most advanced application of AI in the physical world today. Waymo collects massive amounts of data from real roads and billions of miles driven in computer simulations. Manasi Joshi, director of Systems Intelligence and Machine Learning, will explain how Waymo's engineers use these huge datasets to advance how cars "see" the world, predict what other drivers will do, and plan their routes. You'll hear about how Waymo uses ML to train and test these self-driving systems and how its open dataset is motivating further public research in autonomy. This work helps make Waymo's autonomous vehicles safer and more reliable every day in an era of rapid expansion and competition of autonomous ride-hailing services.

Info: DESCRIPTION COMING SOON!

Info: This hands-on workshop explores computer vision techniques for automotive damage detection using the CarDD dataset - the largest public dataset specifically designed for vehicle damage analysis.

Participants will learn how to leverage FiftyOne, a powerful computer vision experimentation platform, to explore, visualize, and build models for detecting six common types of vehicle damage: dents, scratches, cracks, glass shatter, tire flats, and broken lamps.

This workshop is designed for computer vision practitioners, automotive industry professionals, and data scientists interested in applying modern deep learning techniques to vehicle damage detection. Participants will gain practical experience working with a production-grade dataset while learning best practices for model development, evaluation, and deployment.

Info: DESCRIPTION COMING SOON!

Info: Language models have been observed to have emergent capabilities that extend past the distribution of their training data. This technical tutorial explores the practical implementation of a novel dynamic in-context learning (dICL) system within a Graph RAG agentic workflow. We will begin by reviewing the foundational concepts of in-context learning, where Large Language Models (LLMs) adapt to tasks by leveraging examples directly within their input prompts. We will then introduce self-optimizing agents, which extend LLMs with capabilities for real-time and continuous self-improvement through context optimization. While traditional ICL is often static, our approach demonstrates how to create a dynamic feedback loop that enables an agent to learn and adapt in real-time from its own performance.

This system integrates Comet Opik for robust evaluation and LanceDB as a vector-native database for efficient knowledge management.

Info: Productionizing GenAI applications often relies on multi-stage, complex pipelines that manage data ingestion, creation of vector embeddings, vector storage, querying, and response construction. These pipelines need to run reliably, scale efficiently, and recover from failure without manual intervention. Apache Airflow, a leading open-source orchestration framework, provides the structure and flexibility required to operationalize GenAI workloads with production-grade reliability. In this workshop, you'll learn the basics of using Airflow to build and orchestrate end-to-end GenAI workflows.
We’ll explore orchestration through the lens of a Retrieval-Augmented Generation (RAG) use case: a book recommendation system that embeds book descriptions, loads them into a vector database, queries them based on user input, and provides LLM-generated summaries.

Info: DESCRIPTION COMING SOON!

Info: Model Context Protocol (MCP) is an open protocol for AI agent communication that is revolutionizing how AI agents interact with data, and its possibilities for automation and intelligent management erupt when metadata is included. OpenMetadata is an open-source platform that centralize metadata to empower teams to build and manage high-quality data assets through data discovery, observability, and governance. By combining the two, organizations can transform their data operations, realize new ways of efficiency, and unlock the full potential of their data assets.

Join us in this hand-on workshop (or talk) as we:
- Introduce both OpenMetadata and MCP Servers and their impact on a data stack
- Explain architecture motivations for embedding MCP into an open metadata platform instead of using it as a standalone server
- Walkthrough transformational use cases for AI to use metadata to proactively identify and fix data quality issues or enforce access and ownership policies

Info: DESCRIPTION COMING SOON!

Info: High benchmark scores make for great headlines, but they often mask the messy truth of real-world AI performance and adoption. Recent stories like Meta’s LLaMA-4 benchmark controversy, OpenAI’s GPT-4o “sycophancy” issue, and the bumpy GPT-5 rollout show how easy it is for models to look stellar in controlled tests yet stumble with actual users. Benchmarks tend to be narrow, static, and sometimes irrelevant or compromised by data leakage or over-optimization. This talk makes the case that AI evaluation must go beyond leaderboard bragging rights to focus on the realities of each organization’s unique environment.
We’ll explore how and why the gap between benchmark scores and real-world results is widening, drawing on timely case studies and industry trends. Attendees will learn practical methods to create internal benchmarks, stress-test and experiment with models for their own use cases, and monitor them post-deployment to catch behavioral drift, hallucinations, or biases.

Info: Context engineering is the art of dynamically filling an LLM's context window with the right information to help it perform a task to the best of its ability. It's most often the case that when an LLM isn't performing as expected, it's because it doesn't have access to the relevant information needed to succeed.
This becomes an especially challenging task for agentic systems, where conversations can span tens or hundreds of turns. Relevant information can rapidly balloon and exceed context window limits, and outdated context can even sabotage the agent's past progress. As a result, sophisticated systems must be in place to intelligently persist, filter, and evolve context throughout agent runs.
Consequently, context engineering has emerged as one of the most critical skills that AI engineers can develop.

Info: Vibe Coding is changing the way we build software. By combining traditional development with powerful AI tools like ChatGPT, Claude, Gemini, GitHub Copilot, and Lovable, we’re making it easier and faster for anyone to go from idea to execution. This session aims to show how professionals can use these tools to enhance their workflow without losing control of quality, architecture, and security.

It is aimed at developers and tech leads who want to rethink how software gets built. We’ll cover real-world use cases where AI helps you move faster, from automating boilerplate and generating clean components, to speeding up refactoring and improving consistency across teams. More importantly, we’ll discuss how to keep the codebase maintainable and scalable when integrating these tools into professional environments.

By the end of the session, you’ll have a clearer view of how to work with tools like Copilot and ChatGPT not as shortcuts, but as collaborative partners.

Info: DESCRIPTION COMING SOON!

Info: DESCRIPTION COMING SOON!

Info: DESCRIPTION COMING SOON!

Info: DESCRIPTION COMING SOON!

Know what’s Happening Next — before everyone else does.

Get App

Host or Publisher ODSC Team | odsc.com