# Webinar: The hardest problems we solved building Maze

**URL:** https://mazehq.com/event/webinar-the-hardest-problems-we-solved-building-maze
**Date:** 2026-08-26

Have you ever looked at how security gets done and thought, “I know AI could do this better”? Yeah, us too. That’s why we built Maze.

Our CTO and co-founder, Santiago, digs into some of the most interesting, and hardest, problems we solved between a working MVP and the Maze that runs across millions of vulnerabilities weekly:

- Building an accurate context layer that feeds the agents
- Understanding the difference between using one big prompt vs many specialized ones
- Selecting the best model for the job
- Designing an architecture that balances accuracy, efficiency, and cost

It took us over a year, and a lot of engineering creativity, to achieve results that are scalable, accurate, consistent, and won’t break the bank. Come hear about what we learned.

## Speakers

 ![Santiago Castiñeira](https://mazehq.com/wp-content/uploads/2026/08/Santiago-300x300.jpeg)

[ LinkedIn: Santiago Castiñeira ![](https://mazehq.com/wp-content/themes/okd/static/img/linkedin.svg)](https://www.linkedin.com/in/santiagocastineira/)

 Santiago Castiñeira Co-Founder & CTO, Maze 

Co-Founder and CTO of Maze, the agentic reasoning platform for vulnerability management, where he leads engineering. He brings years of hands-on experience in the field and spent 15+ years building and scaling software in senior technical roles at Amazon, S&P Global, Monad Inc, and Jungle Robotics. He pairs deep engineering expertise with a track record of turning complex security challenges into products teams can use.