AI Track at Web Summer Camp 2025: Learn by Doing with Practical, Production-Ready AI Tools
It’s mid-2025, and introducing an AI track to Web Summer Camp feels not just timely, but necessary. However, we didn’t want to do it just because everyone else is doing it. We wanted to do it right. That means keeping our signature hands-on format, focusing on knowledge you can use immediately, and delivering a meaningful learning experience.
At Web Summer Camp, each track is centered around immersive, hands-on workshops—and the AI track is no exception. Designed for practical learning, these sessions will equip participants with real-world skills and actionable insights.
Here’s what’s in store for those diving into the AI track:
Reasoning in LLMs by Marko Velić, Engineering Manager and Machine Learning Researcher @ Google
Learn how to build large language models with "reasoning" capabilities. This workshop demystifies how today’s "thinking" models like DeepSeek, Gemini, and Grok3 are trained and fine-tuned. You’ll start with a base model lacking math skills, then train it using supervised fine-tuning (SFT) and reinforcement learning to become capable of solving math word problems. The goal? Leave the workshop with your own fine-tuned reasoning LLM and a deep understanding of what makes these models tick.
Serving LLMs from the First Principles by Marijan Smetko, Software Engineer @ Google
Get your hands dirty and build two AI-serving systems from scratch. The first will rely on core tools like PyTorch and FastAPI to teach you foundational model serving concepts. The second will introduce you to Nvidia Triton and show how to create a performant, production-ready AI inference system. If you're aiming to deploy LLMs or any AI model at scale, this one is for you.
Solving Complex Tasks Using AI Agents by Davor Runje, Head of OSS Engineering @ AG2
Explore AG2, a powerful framework for building collaborative AI agents. This workshop will showcase multi-agent orchestration, tool integration, and human-in-the-loop workflows. With real-world examples, you'll see how coordinated agents can tackle intricate tasks, offering insights into building scalable, intelligent systems.
Development of AI Multi-Agent Systems Using Python and LangGraph by Ive Botunac, AI Research Scientist & Engineer @ Alfatec
Dive into the LangGraph framework and learn how to design modular multi-agent systems. You'll define states, nodes, and transitions to create dynamic, responsive AI architectures. This is a great fit for developers interested in scalable AI systems capable of real-time interaction and coordination.
Developing Production-Ready Apps in Collaboration with AI Agents by Alex Shershebnev, Head of ML/DevOps @ Zencoder
Join this interactive session to build a production-ready Python app alongside Zencoder—an AI coding agent platform. Learn best practices for working with AI agents, from setting up your environment to integrating real-time collaboration in code. Bring your laptop and your IDE of choice (VSCode or PyCharm recommended), and walk away with a working app and practical experience of what it's like to develop side-by-side with an AI assistant.
Additionally to workshops, the program also includes talks:
Client side Web AI Agents: Building smarter user experiences for a future agentic internet by Jason Mayes, Web AI Lead @ Google
Explore how generative AI models running entirely on the client side can transform the user experience on modern websites. This talk highlights how JavaScript-based Web AI can deliver agentic behaviors without relying on cloud infrastructure. Discover practical applications that enhance speed, autonomy, and competitiveness in user-facing features.
Future of the future: creating an AI product beyond hype, by Radovan Bacovic, Staff Data Engineer @ GitLab
Gain insights from the development of GitLab Duo, an AI product built quickly with a strong focus on user experience. This talk shares practical lessons from iterative development, internal testing, and real-world product strategy.
Whether you're an experienced engineer looking to deepen your AI chops, or a developer curious about how to bring LLMs and agents into your workflow, join us in July for three days of learning!