I identified that our bug-reporting system was in drastic need of a revamp and decided
to do something about it. I proposed changes to add structure and clarity,
and I got buy-in from stakeholders throughout the organization.
I wrote this blog post
describing how we changed our chat-ops processes, improved our response times,
and made our stakeholders happier.
In mid-2023, I began raising concerns about our practice of deriving custom credit attributes
based on raw credit-report data. It was slow, error-prone, carried increased compliance risk, and did
not follow standard industry practices. Working with our VP of Data Science, our
Credit Risk department, our Compliance department, and other key engineering stakeholders, I crafted
a proposal for shifting to precomputed credit attributes provided by the credit bureau.
(I also established a back-up plan that would allow us to decrease the time it took to create new
custom attributes from weeks to minutes, using easily updatable rule sets and a lightweight parsing
library based on JSONata).
Fortunately, though, the proposal was adopted with enthusiasm from our C-suite, and we began contract
negotiations with the credit bureau. While those negotiations were protracted, we were eventually
able to begin our integration in early 2024. All future ML models will rely on these precomputed
attributes, rather than our in-house derived attributes, which will greatly speed up our model
development lifecycle.
Led an initiative to apply the Black auto-formatter to Python code throughout the codebase, via pre-commit hooks.
Created an engineering onboarding hub, Octane University, and promulgated the use of C4 system diagrams.
Led an initiative to significantly reduce decisioning speed. Within the span of six months, we reduced our latency from a p95 of nearly 120 seconds to under 30 seconds.
Our Encantos Cloud team created serverless resources using AWS SAM.
This allowed us to manage our own infrastructure as code and rapidly iterate without the need for additional DevOps resources.
Experimenting With Kids: 50 Amazing Science Projects You Can Perform on Your Child Ages 2-5 was published in April 2020.
Experiments for Newlyweds: 50 Amazing Science Projects You Can Perform With Your Spouse was published by Sourcebooks in April 2019.
Presented a session at the Linux Foundation's annual API Strategy and Practice Conference in September 2018.
The title of the session was, "The Evolving API: Designing While Requirements Are Still In Flux."
Experimenting With Babies: 50 Amazing Science Projects You Can Perform on Your Kid was published by Penguin Random House in October 2013. It has since sold more than 100,000 copies.