Daytona AI Builders SF March '25 @GitHub
An event dedicated to exploring all things AI Engineering!
Agenda
🕒 5:30 pm – 5:35 pm
Welcome and Opening Remarks
🎤 Marijan Cipcic, Principal Events Manager at Daytona
🕒 5:35 pm - 6:00 pm
Keynote "Building reliable web agents in production"
🎤 Paul Klein IV, Founder of Browserbase
Outline:
An agent is only as good as its tools. If you’re building an agent that requires web browsing capabilities, you may be facing issues with reliability. In this talk, Paul Klein (founder of Browserbase), will share the common practices his customers use to build reliable web agents in production.
🕒 6:00 pm - 6:20 pm
Keynote "Building Infrastructure for AI Agents"
🎤 Ivan Burazin, Co-Founder and CEO at Daytona
Outline:
AI coding agents today are constrained by primitive infrastructure, limited to basic file operations with no access to runtime or output context—resulting in poor outcomes as agents can only assume their suggestions are correct. This talk will dive into the challenges agents face due to these limitations, explore how we can overcome them, and demonstrate how this unlocks the true potential of AI coding agents.
🕒 6:20 pm - 6:40 pm
Keynote "Beyond Autocompletion: Engineering and Optimizing AI Agents for Software Engineering"
🎤 Kilian Lieret Ph.D., Research Software Engineer at Princeton University
Outline:
Language models have moved beyond code completion and are increasingly being used to tackle software engineering tasks in a more autonomous, agentic way. This talk introduces SWE-agent, a straightforward yet powerful open-source agent framework designed for software engineering but adaptable to a variety of domains. After a quick live demonstration, I will dive into the engineering challenges of building agents, particularly in enabling them to execute code and interact with sandboxed environments. I will also discuss lessons learned from optimizing the SWE-agent prompts, tools, and the agent scaffolding for software engineering tasks.
🕒 6:40 pm – 7:00 pm
Keynote "Cloud Sandboxes for AI Code Execution"
🎤 Toma Puljak, Software Engineer & Developer Advocate at Daytona
Outline:
As AI applications grow in complexity, the need for secure, isolated, and high-performance environments to execute AI agent code becomes critical. This talk explores the concept of cloud sandboxes—isolated, ephemeral environments designed for running AI workloads. We’ll dive into key use cases, such as testing, debugging, and fine-tuning AI agents, and discuss why speed and isolation are essential for these workflows. The session will also cover the architectural components of a cloud sandbox, from provisioning and resource allocation to execution and teardown.
🕒 7:00 pm - 8:30 pm
Networking
With snacks and beverages
_________________________________________________________
About event
This is dynamic gathering for AI enthusiasts, innovators, and professionals to collaborate, share ideas, and explore the latest advancements in artificial intelligence. Whether you're building AI products, researching cutting-edge algorithms, or simply passionate about the field, join us to connect, learn, and drive the future of AI forward.