5 Questions With Our New Machine Learning Engineer: Javen Xu

By Jason Kapler


Artificial intelligence is what powers EvolutionIQ’s claim guidance system. And it’s our machine learning engineers who develop the algorithms and models that enable the technology to analyze, reason, and understand just like a human would. Staff Machine Learning (ML) Engineer Javen Xu joined EvolutionIQ in early 2023 and is part of the ML team that focuses on our Long-Term Disability product modules (LTD). With degrees from University of California, Berkeley and Stanford University, Javen’s previous roles include quantitative research at financial giant BlackRock, co-founding Cardinal Robotics, and machine learning at Lyft. 

What is your primary focus as a machine learning engineer at EvolutionIQ?

We are usually focused either on client delivery – which is getting the product ready for launch mode – or we are developing new modules or new features. Usually the beginning phase of a client project is a lot of planning, creating technical design documents, and getting feedback from the client that we need so that we can ensure there is alignment. After this it’s a lot of coding and coordinating with the product team, the data team, and the applications team to iterate on the different versions of code we create and then calibrate the data. While EvolutionIQ makes its AI products scalable, every client will still have unique data and a unique set of data challenges. There will always be some piece that needs attention. Although 80 percent of the work we do in building our models is scalable across a number of deployments – so we’re not starting from scratch each time – the platform is not just plug-and-play. It still usually takes some type of iteration for each specific client. It’s not customization. Rather, it’s specialization so that we ensure our model works with their unique data and that it understands how their internal systems work.   

What attracted you to EvolutionIQ?

I wasn’t familiar with insurance, but the business model of EvolutionIQ really stood out to me. What the company does makes so much sense and EvolutionIQ has a very real mission. And the value that the company creates is really a win-win – for the claimants who get back to work and it’s a win for the insurance carriers. The business model is very concrete. I understand it. And it connects very well with my machine learning/AI background. EvolutionIQ has created something new that provides a real service and that is very appealing to me. 

What interests you most about artificial intelligence? 

AI is a very exciting technology. A lot of things can be improved by intelligent systems – like helping people do their jobs better, in our case, with claims guidance. What’s important to me is applying AI in a way in which I can understand the specific value proposition – so that it’s very real, and that it makes sense. Because I can understand the value of the things we’re building here at EvolutionIQ, I can make an impact and I can see how that impact is happening. The way we marry our technology to business application is very important to me.

What’s it like working on your team?

There is no wasted work. We definitely have a fast speed and we iterate fast – so there’s no wasted time, which I like. For every task we do, we plan it well and we know at the end of one month or two months we’re going to achieve something – either with a new client being successful or in developing and launching an entirely new AI module or product feature. We know the work is going to be useful when it’s finally in the hands of a client. I also like the fact that I’ve had opportunities to embrace new challenges and venture into new technical areas beyond my initial expertise here at EvolutionIQ. Recently, I began leading a project focused on enhancing our understanding of claim notes and actions. This endeavor goes beyond my familiar domain of machine learning – requiring me to delve into the intricacies of data pipelines and the application side of things. This multifaceted project has served as an immensely valuable learning experience, allowing me to evolve into a more versatile engineer with an aptitude for conceptualizing larger system designs.

How do customers respond to your work?

Recently, I had the opportunity to play a pivotal role in delivering a significant project for an esteemed insurance carrier that already had its own internal data science and analytics team. Their team had been working on a similar solution to our product, which initially led to some skepticism on their part regarding the additional value we could bring to their existing systems. To address these concerns, we conducted a pilot where we had their insurance examiners utilize our systems alongside their internal tools. The results were truly remarkable. Both the subjective feedback from the examiners and the objective metrics and numbers painted a highly favorable picture of our solution. The head of the client’s data science team, who initially held reservations, went from a skeptic to a champion for our product. As the lead for machine learning on this client engagement, I felt an immense sense of pride in my work. This successful delivery for the client reinforced the value our product brings to the table – and it showcased our ability to surpass expectations and win over skeptics.

Hear from our other team members:

previous articles