AI DevSummit 2025 + DeveloperWeek Leadership 2025

Abhai Pratap Singh, Amazon, Senior Product Manager-Technical

"From Technical Possibility to User Value: Product Management Strategies for Successful AI Implementation" explores the delicate balance between advancing AI capabilities and maintaining genuine user value. Drawing from experience leading Alexa's voice assistant development, this session reveals practical strategies for putting user needs at the center of AI product development. Learn how to avoid the trap of 'tech for tech's sake' while driving meaningful innovation. Through real-world examples from voice AI and multi-modal interactions, discover frameworks for evaluating AI features, measuring user impact, and making strategic decisions that enhance rather than complicate the user experience. Perfect for product leaders navigating the AI revolution.

Mark Fussell, Diagrid, CEO
Yaron Schneider, Diagrid, Co-Founder / CTO

As developers push the boundaries of AI-driven automation, the challenge of orchestrating and managing autonomous agents at scale becomes increasingly complex. Dapr Agents is an open-source framework that dramatically lowers the creation of production ready agentic systems, powered by the CNCF Dapr's (dapr.io) distributed application runtime, used by thousands of enterprises today in mission critical services. By combining LLM-driven reasoning with Dapr’s virtual actor model, pub/sub messaging, and stateful workflows, Dapr Agents enables developers to build intelligent, scalable, and fault-tolerant multi-agent systems.

This talk will provide a deep dive into Dapr Agents, demonstrating how it facilitates structured LLM interactions, tool selection, memory retention, and event-driven communication. We'll explore practical use cases, from task automation to collaborative agentic workflows, and discuss best practices for designing real world, robust agent-based architectures. We will also compare different agentic frameworks in use today and show how many of them only address a few of the requirements needed to go into production environments. Attendees will gain hands-on insights into implementing Dapr Agents for real-world applications, optimizing performance, and seamlessly integrating with existing cloud-native infrastructures.

If you're interested in building intelligent, scalable, and resilient agentic systems, this session will equip you with the knowledge to easily build these.

Rob Whiteley, Coder, CEO

With great power comes great responsibility—and yes, coding agents can backfire if they’re not thoughtfully deployed. Instead of boosting productivity, they can frustrate developers, slow down progress, and even diminish your competitive edge.

Join Coder CEO, Rob Whiteley, as he unpacks three essential strategies organizations should embrace when scaling agentic coding agents enterprise-wide. Drawing from firsthand lessons and internal experiments—where agents effectively resolved nearly half the tasks we threw at them—we’ll share real-world scenarios showcasing engineering work AI agents excelled at. We’ll also highlight where they fell short, detailing what went wrong and what we’d do
differently next time.

You’ll walk away understanding:

* How realistic expectations ensure AI boosts rather than hinders productivity.
* Why effective human-agent collaboration is critical for developer satisfaction and innovation.
* How clear governance frameworks can prevent technical debt and keep productivity flowing.

Learn from our experiences and ensure your organization’s deployment of AI coding agents empowers your developers, ignites innovation, and solidifies your market leadership.

Sri Harsha Yayi, intuit, Staff Product Manager
Vigith Maurice, Intuit, Principal Engineer

At Intuit, the ML teams encountered challenges when managing and executing inference on high-throughput streaming data. Integrating with diverse messaging systems like Kafka, Pulsar, and SQS required considerable effort, while enabling intermediate data transformations and inference added complexity. Additionally, dynamically scaling inference workflows to handle fluctuating event volumes posed unique challenges.

To address these issues, we developed Numaflow, an open-source, Kubernetes-native platform designed for scalable event processing. Numaflow simplifies connections to various event sources, empowers teams to process and infer on streaming data with ease, and integrates effortlessly with existing infrastructure. This session is ideal for ML engineers, data scientists, and anyone interested in asynchronous inference on streaming data. We’ll dive into how Numaflow eliminates bottlenecks and optimizes inference on streaming data.
 

Shomron Jacob, Iterate.ai, Head of Applied Machine Learning & Platform

This AIDev Summit session will navigate an increasingly pivotal crossroads: the decision between investing in proprietary, custom-tailored Large Language Models (LLM) or capitalizing on the versatility and ease of public generative AI services.

The session will begin by demystifying the complexities of private LLMs. With domain-specific capabilities and enhanced data security, these models have faster customization and compliance with industry-specific regulations. Yet, they also pose challenges: a bigger investment, infrastructure requirements, and ongoing maintenance. These elements necessitate a thorough examination.

Next, the session will scrutinize public generative AI services, exploring the inherent benefits of these ready-to-use solutions. With their scalability, diverse applications, and lower upfront costs, they hold significant appeal. But they also come with their own set of considerations, such as data privacy, standardized performance, and reduced control over the model’s behavior.
With real-world examples, we will walk through how various organizations have approached this decision, the results they achieved, and the invaluable lessons learned.

The session will then go into a decision-making framework, with the purpose of enabling attendees to assess their options between private LLMs and public generative AI services more effectively.