Work

I lead the Integrations team at Regie.ai. We connect Salesforce and HubSpot to Regie so customers can bring their data in and send updates back. I’ve recently been working on a monday.com integration too.

The whole sync architecture is fascinating to me. An import stops halfway through. Where do you pick it up again? How do you know an update actually made it to the other side? I design how the first import and later updates work, including how jobs recover when permissions change or a provider’s API does something unexpected. Each customer’s data has to stay separate through all of it.

I also spend time with customers to understand how they actually use their CRM. With Commvault, I worked with their leadership from onboarding through renewal. We went through how leads came in, who owned them and where automation handed things over to a person. Those conversations helped us figure out what we needed to change in the product and what their team needed help with.

At Nymble, I worked on a cooking robot. My part was turning a recipe into instructions the machine could follow: what to dispense, how much and when, along with the cooking steps.

Some of those actions happened at the same time, so we had to coordinate the timing. Even skipping a dispense needed an instruction. You had to tell the robot to do nothing there.

1. RecipeIngredients and cooking steps
2. InstructionsQuantities, timing and order
3. ExecutionWhat the robot actually did
4. ComparisonDid it follow the instructions?

I also designed the instruction logs and a test bench so we could compare what we’d asked the robot to do with what it actually did. When those didn’t match, the logs gave us somewhere to start looking.

My first full-time job was at Incubyte, working on lending software for Republic Finance, a US consumer lender. We used Java with Micronaut and Go with Gin.

We did a lot of pair programming and wrote tests before the implementation. Working through code with another person taught me a lot about making it readable. I still pay attention to that when I review code now.

During my internship at Siemens, I worked on a question: could we use reinforcement learning to reduce a building’s energy use while keeping the rooms comfortable? I worked on a controller for heating and cooling, with comfort limits defined by ASHRAE.

There was also the work of getting models into containers, running them remotely and validating the existing ones. The project later became a challenge at the Siemens Global Hackathon 2020.