Meet a Text IQer: Senior R&D Engineer Weichen Wang
Company

Meet a Text IQer: Senior R&D Engineer Weichen Wang

In our first installment of this new series, we interviewed Weichen Wang, who holds a Master of Mathematics from the University of Waterloo and has previously worked as an engineer at enterprise software companies like Zenefits and Clio. Read on to learn what it’s like taking on one of the unsolved problems of AI.

In our first installment of this new series, we interviewed Weichen Wang, who holds a Master of Mathematics from the University of Waterloo and has previously worked as an engineer at enterprise software companies like Zenefits and Clio. Read on to learn what it’s like taking on one of the unsolved problems of AI.

How did this whole thing begin?

I came home from school one day in the fourth grade and there was a PC in my bedroom. I didn’t even know what a PC was at the time. Later I learned about the sacrifice behind it. My parents worked in a state-controlled chemical plant in China. We were working class, earning a few hundred dollars a month. And that PC cost a few thousand.

It was a huge investment and completely my father’s decision. He got the idea after he read The Road Ahead by Bill Gates. The book talked about the internet in fancy terms that didn’t exist yet. My father thought that maybe computers were the future, and that maybe I could be a part of that future.

What I really admire and appreciate is that he never forced the idea on me. It was important to him that there was no command. He just put the computer in my room and left it there, waiting for me to discover on my own. 

Where does your passion for programming come from?

I fell in love with programming because I could express myself. Writing something in code can have an actual, real-world impact. If you write a book, you send it off and hope people absorb it. But if you write a program, you can run it, and it does things for you. You can actually materialize ideas.

I remember the first non-trivial program I wrote. It simulated lottery tickets. You entered numbers, bought tickets, rolled the dice, and it told you whether or not you won. And the first thing I did was share it with my dad. He was hyped, but also a little disappointed that it involved gambling.

He still told me to keep going. I guess my love of programming comes from a combination of things: I was good at it, I was proud of it, and there was recognition, and all these things gathered momentum and kept building on each other.

Why did you choose to join Text IQ?

I’ve had a few different jobs since I graduated college. None of them has excited me like this one. I’m drawn to AI for sensitive information. The challenges are big and the possibilities are infinite. We’re dealing with massive amounts of data, and we’re applying unsupervised machine learning to it. We’re at the intersection of two state-of-the-art fields: big data and AI.

We’re working with some incredibly cool techniques, because we’re taking on one of the unsolved problems in AI: how to ingest data in never-before-seen structures, and induce structure into it, so that it can understand the underlying meaning, people, and relationships.

Because the data we’re dealing with so huge, and unstructured, we work with interesting tools, like Elasticsearch, Apache Beam, and Redis. But where things get interesting is how we assemble these tools into a scalable, distributed architecture.

And the implications of what we’re doing are big. Clients come in with hard deadlines that they’re desperate to meet. These companies and agencies are some of the largest entities out there, and they’re depending on us to navigate serious, high-stakes situations.

What were you first conversations like?

I was struck by Apoorv, our CEO. First, I just thought it was very cool that he had a Computer Science PhD in Columbia, working with some of the brightest AI researchers out there, like John Paisley. And when we spoke, I found that he had a clear picture of what he wants the company to be. When I was asking him technical questions, I could sense that spark, that engagement, and I felt a connection.

I told Apoorv that I wanted my work to matter. I wanted to solve critical needs and make a business impact. And I didn’t want to blend too much with others. I wanted to make an impact that could be traced back to me as an individual. I also told him that I wanted to ”get shit done,” and work on cool things.

Apoorv told me that this company was about finding a needle in a haystack. This is a core problem. Yes, we have several projects in specific domains, like helping out litigations, compliance events, and privacy concerns. But at the bottom of all of our projects, there’s a tremendously interesting category: finding sensitivity in the artifacts of our human interactions.

What’s the coolest project you’ve worked on so far?

We have a project right now that represents a new order of magnitude of data volume. The way I see it is that this sort of scale is also a feature. A restaurant and a hotel might both serve food, but because of their different sizes, the design is inherently different. Every order of magnitude change requires a new approach. That means we’re evolving quickly, because we’re continuing to scale up, and we’re redesigning things so that they stay functional, robust, and reliable for larger and larger sets of data. I love diving into these oceans of code and making reason out of it.

And where things get really interesting is when we receive information in a structure that we’ve never seen before: not just emails, but chat logs, calendar invites, notes in various forms. The list is not literally infinite, but it practically is. Just consider that really there are as many structures as there are discrete communications, because humans interact in an organic way, and language is deeply ambiguous and dependent on context.

For example, we detect codewords, so we know that “reuben sandwich” in this record isn’t a sandwich, but a restricted stock. And we’re detecting this using a very sophisticated system. Pretty cool.