author profile pic

By Eleanor Bennett

Interview

5 min read

In the latest instalment of our interviews speaking to leaders throughout the world of tech, we’ve welcomed Ricardo Michel Reyes, Chief Science Officer at Erudit AI.

Tell us a bit more about the business you represent.

Erudit is an AI web software that makes sense of what’s going on in organizations every day. It provides metrics and insights on the more ambiguous parts of the business: its people.

Companies normally understand their employees through productivity tools, which are one-dimensional. Erudit reveals the complexity of teams based on organic textual data and helps you understand them through updated, even daily metrics. It lets you understand your people on a deeper level, without relying on surveys.

What is it like to go from being the CTO to the CSO? What does that journey entail?

The CTO normally doesn’t program or code, they manage the team. He/she should be very balanced, and a good listener. Also, the CTO gets the team to solve things that we already know how to solve.

I would love to be able to do that but Erudit requires me to code a lot of neural networks. As CSO, I code a lot. More than that, I have to understand the technology and the specific problems we try to solve from the scientific perspective. A CSO also has to be constantly in the know of the latest technologies and studies, as opposed to existing best practices in leadership and tech development.

In the beginning, I was both CTO and CSO. But now that we’re both improving the current Erudit platform and also developing new features and solving new problems, it’s harder to do both! Now we have a very good CTO that works well with the team while I can focus on keeping Erudit cutting-edge.

Being the CSO, what do your day-to-day responsibilities look like?

Keeping Erudit state-of-the-art! It entails reading an awful lot. There are websites with code that report the newest technologies in the different realms of AI every day. We cover a lot of different Natural Language Processing tasks, not just one, so it’s important to read up on how it’s changing and developing.

We have to keep up-to-date with the architecture and data sets out there and know what other people are doing, so we know how to put ourselves on the same level or go further. There’s https://arxiv.org/ and https://scholar.archive.org/ where you can find research papers from different universities.

Being a CSO is looking into the future and a lot of interdisciplinary things. You have to know a lot about a lot of things so you can put it all together when finding a solution. The work is, basically, the Product Team proposes a problem, then we see if AI can solve the problem: if it’s even possible, if it makes sense psychologically if we can support our claims statistically. Then if it’s possible, we think of how: through the lens of psychology then how to model it mathematically. Then it’s handling the team to conduct experiments on the different parts of the solution so they can be built and integrated by the CTO. We pretty much do the research part, the MVP part of what the CTO is going to build in a long-term, scalable way.

Can you tell us more about your work with the European Space Agency?

I’ve been working with them through several partners or companies in various European countries for a little over a year. What we’re working on is quantum computing applied to satellites. Satellite imaging is huge! You have terabytes and terabytes of data. So to be able to process all that data, you need either a supercomputer, a lot of GPUs, or in this case quantum computing.

Satellites capture light in different spectrums that we cannot see. We normally see RGB; satellites have bands, which are the different parts of the electromagnetic spectrum they capture. To give you an idea of their capability, humans see, like, 3 bands of a satellite but there are some with 150! They see a lot of things we cannot see: different types of molecules, the quantities of the molecules on earth, and on a very large scale.

It’s useful for climate change monitoring, disaster prevention, assessing how mining is affecting or polluting areas, agriculture, forestry… The technology can be applied to taking care of natural resources and studying different parts of the world (because you can be in Spain and study China without having to ask them permission, for example).

It’s a new frontier. It’s quantum computing with AI with satellite imaging–all at the same time!

What do you see as the hottest trends within your industry today?

Transformer Neural Network… How can I explain this? How Science was done before is that you perform experiments then you try to reach an equation, like for example: speed is distance over time. You could win a Nobel prize for an equation! But now, with so much data and bigger, more ambiguous, abstract problems, you cannot just explain it in an equation. Or if you do, it’s after you’ve approximated the problem in many different empirical ways. So most of the trends now have to do a lot with the attention mechanism and how you can use it to make your models better.

We experience this as humans! Your eyes and ears capture light and sound constantly, but you don’t hear all the sound nor see everything. That’s the attention mechanism. How do you focus on one signal or just a part of the signal?

There’s a lot of quantum computing involved, which is a very important trend. Quantum computers have been around for 7 years, not really for public applications. But it’s something that will have applications in 25 years or so. Imagine we started computers in the 40s, and now we’re here! It will be the same for quantum computers in the future.

Can you tell us what book you are reading currently?

The Dawn of Everything by David Graeber and David Wengrow who are trying to retell the story of humanity from the latest research. When you go to school, they teach you a narrative linked to your country. I’m from Mexico, they teach you about the world from the Mexican perspective. But when you just base learning on the data, the story is different.

So in the book, they reevaluate the story of marriage, agriculture, private property, slavery, war, religions, etc. They tackle common threads of humanity across cultures and how they occur. Government, how monarchies and feudalism have occurred across cultures and how people organized themselves.

Do your technical teams or do you use log analysis as part of your role? If you do, how do you find this helps day-to-day operations?

I can’t imagine life without log analysis because otherwise, it will be hard to know what’s wrong. Everything the pipeline does goes to the log. If one day, there’s no data the first thing we look at is the log to see if it got processed or if something got stuck. Everything we do in Erudit is logged.

What can we hope to see from your business in the future?

I think we’ve now reached a good understanding of the customer: people leaders. We know who they are, what they want, and how they want it.

Now I’m thinking about how to create something that adds value to the employee and not just the manager. It would be cool to create a well-being tool for employees that empowers them to take steps to live better, healthier, and happier.

If you enjoyed this post and want to keep reading our best articles then why not check out our article on Prometheus tools or our guide to opensearch vs Elasticsearch?

Get the latest elastic Stack & logging resources when you subscribe

backReturn to Blog

© 2022 Logit.io Ltd, All rights reserved.