Kavak x OpenAI Hackathon: Self-Improving AI Systems
- Date
- 2025-10-23
- Location
- Ciudad de México, Ciudad de México, Mexico
- Host
- Mexico Tech Week 2025
About this event
If you care about AI systems that can do more than complete a prompt—systems that observe, adapt, evaluate themselves, and get better over time—this hackathon is built for you. Kavak x OpenAI Hackathon: Self-Improving AI Systems brings builders together in Mexico City for a focused day of hands-on experimentation around autonomy, iteration, and practical AI product design. About the Event This is an in-person hackathon for people who want to move beyond static demos and explore what it takes to build AI that can refine its own behavior. The theme, Self-Improving AI Systems, points to a very specific challenge: how do we design systems that learn from outcomes, adjust their strategy, and become more useful with repeated use? You can expect a builder-first environment where the emphasis is on making, testing, and sharing real work. Whether your background is in engineering, product, research, or applied AI, the format is designed to help you turn an idea into something concrete. The collaboration between Kavak and OpenAI signals a practical, forward-looking focus on how autonomous systems can be developed in ways that are both technically ambitious and grounded in real use cases. This is also a community event in the best sense of the word. You will be in the room with other people who are actively thinking about agents, evaluation loops, tool use, orchestration, and the next generation of AI products. That mix of technical curiosity and shared momentum is part of what makes a hackathon like this valuable. What to Expect The day starts early, at 7:30 AM CST, and the structure is geared toward giving teams enough time to move from concept to prototype. While every hackathon evolves around the energy in the room, you should expect a rhythm that supports both focused building and useful interaction with other participants. A typical flow for an event like this includes: Check-in and kickoff to get aligned on the day, the theme, and the working format Team formation or project setup for attendees arriving with an idea, a team, or an interest in joining one Dedicated building time to prototype self-improving AI workflows, agents, or evaluation systems Live collaboration with peers who can pressure-test your assumptions and help sharpen your approach Project sharing or demos at the end, where teams present what they built and how they approached the problem The strongest hackathon projects usually do more than showcase a model response. They demonstrate a loop: setting goals, taking actions, measuring outcomes, and improving based on feedback. That could mean experimenting with memory, evaluators, decision policies, retrievers, tooling, or multi-step orchestration. It could also mean exploring the product layer: how users interact with a system that changes over time, and what trust, control, and transparency should look like in that experience. You should come ready to think in systems, not just prompts. The event theme rewards projects that can show how an AI application becomes more effective through iteration rather than one-off performance. Why Attend If you have been wanting a reason to stop reading about autonomous systems and actually build one, this is that reason. A good hackathon compresses weeks of scattered experimentation into a single day of concentrated progress. The in-person setting adds speed: fewer delays, faster feedback, and more opportunities to solve problems with the people around you. There is also a clear advantage in working on a tightly framed theme. Self-improving AI systems is broad enough to invite creativity, but specific enough to push better thinking. Instead of making another generic AI app, you will be challenged to answer harder questions: What should the system optimize for? How does it know whether it improved? What signals matter? Where do autonomy and oversight meet? You may leave with: A working prototype or proof of concept A sharper understanding of agentic system design New collaborators from the Mexico City AI and tech community Practical ideas for evaluation, iteration, and product architecture Better instincts for what makes autonomous AI useful in the real world For founders, engineers, and researchers alike, the value is not only in what you build during the event. It is also in the conversations that happen around the build: what approaches are emerging, what patterns are holding up, and where the field still feels open. Practical Details This event is in person in Ciudad de México, Mexico, which means you should plan for a full on-site working session rather than a casual drop-in. If you do your best work with a laptop, a clear idea, and a room full of serious builders, this format will suit you well. The start time is Thursday, October 23 at 7:30 AM CST. Because the day begins early, it is worth planning your morning in advance and arriving ready to get started. Hackathons move quickly at the beginning, and the teams that gain momentum early often make the most of the time available. A few useful ways to prepare: Bring your laptop, charger, and anything you need for a full day of building Come with a project idea, a technical question, or at least a theme you want to explore Be ready to collaborate if you are not arriving with a full team Think ahead about how you would demonstrate improvement, evaluation, or adaptation in your project Most of all, come ready to build something specific. The people who get the most out of this kind of event are the ones who show up with curiosity, technical ambition, and a willingness to test ideas in public.
Who should attend
This is for people who want to build AI systems that act, learn, and improve—not just generate outputs. - **You are an engineer or developer** excited by agents, tool use, evaluation loops, memory, orchestration, or other building blocks of autonomous systems. - **You are a product builder or founder** thinking about how AI can create compounding value through feedback, iteration, and better decisions over time. - **You are an AI researcher or practitioner** who wants to explore practical applications of self-improving systems in a fast, collaborative setting. - **You like learning by making** and would rather spend a day prototyping, testing, and discussing tradeoffs than just watching talks about the future of AI. - **You want to meet the local AI and tech community in Mexico City** and connect with other ambitious people working on applied autonomy. - **You do not need to have everything figured out before you arrive**—if you have strong curiosity, useful skills, and an interest in the theme, you will have plenty to contribute.