May Walter

What is the most significant initiative you have led recently?
I built the technology that underpins Hud. When we started, everyone told us it was impossible to trace every function in production without slowing down the system or paying a fortune for the data. That was exactly what made me want to solve the problem. Today, Hud runs through an SDK that takes just a minute to install and operates across millions of services at our customers' organizations.
This year, we took it a step further. The sensor learns how each function behaves under normal conditions. When a code change is introduced, it can identify the impact of that change, often before users notice anything.

What do you consider the company's key strengths?
I come from a research background, where you learn to think in reverse. When investigating a system, you need to move through it undetected and extract only what matters. We applied the same approach to observability. Most of the time, the sensor remains almost silent, sending only a small amount of summarized data. When something breaks, it sends the full picture: which code ran, what data it processed, and the state of the machine at that moment.
This allows us to see every function, including the smaller ones that other tools miss. Beyond the technology, we have a team of people who chose to tackle a difficult problem at a layer most engineers prefer to avoid. That is no small thing.

What key initiatives have been undertaken recently to strengthen the company in the coming years?
We decided not to build another dashboard. AI agents now write a substantial portion of the code, so we connected Hud directly to them, including Cursor, Claude Code, and Kiro. An agent modifying a function can see how it actually behaves before making changes and receive all the evidence it needs when something goes wrong.
We also introduced checks that compare every pull request (PR) against actual production behavior, with these checks running in CI as well. To me, this is the most important change we have made this year, because it shifts Hud from responding after an incident to preventing problems before they occur.

What are the company's main challenges in the coming years?
Performance and reliability are my biggest concerns. The sensor runs inside our customers' most critical systems, and as those systems grow, the bar keeps rising. We have no room for error.
Then there is the pace of change. Tools and agents evolve every few months, and we need to integrate with all of them without becoming dependent on any single one. Another challenge is recruiting the right talent. People with deep runtime expertise are rare, and we compete for them with the world's largest companies. I invest a great deal of time in this, and every minute is worth it.

What advice would you give to a manager at the beginning of their career?
Stay technical. I've seen many excellent managers distance themselves from the code as soon as they take on a team, only to find themselves a year later unable to ask the right questions. In the AI era, that makes even less sense.
Another lesson I learned through research is that most of the work goes into understanding what is really happening, not finding the solution. New managers tend to rush into solving problems. It is worth pausing to make sure you truly understand the problem first. And hire people who challenge you. A team that agrees with everything you say simply isn't telling you what it sees.

What is important to remember during difficult times?
In research, most attempts fail. You can spend weeks pursuing one approach only to discover it leads nowhere, and this happens again and again. What keeps you going is not a positive mood but a method: one more hypothesis, one more experiment, one more day. Every experiment should result in either a win or a lesson, and both count as success.
The same applies to running a company. When things get difficult, I return to the fundamental questions: What do we know for certain, and what is the next step? I also share the full picture with the team. People cope far better with a problem they understand than with silence.