After more than 15 years at Cars.com I had some hints that the product was facing big headwinds and my time there was probably coming to an end.
I was gracefully given time to close out before my last day. My final task was to decommission entire NewCars.com, business unit’s tech. All of it is no nonexistent! It was the last of many valuable learning experience. And it gave me a chance to reflect on everything I had accomplished.
During those years I touched a huge sprawl of code repos written in many languages across racks and instances, legacy, and modern. I learned the entire two sided automotive business supported in the middle by the Lead System and the core: “PingPost”. Starting as a feature builder with a team of 20, I ultimately became technical point person of a team of 3! Even so, we never really needed to be on-call. The system maintained stability throughout and the revenue kept flowing. I improved reliability, refactoring and decrufting in PHP, and Java, built new features and ETLs in Python, pushed for awesome tools like Slack, LucidChart, Tableau, Retool, before they were mainstream, documented and preserved institutional knowledge, modernized infrastructure, containerized legacy applications, and created roadmaps to upgrade with stability with core systems created decades ago. I scanned the whole enterprise looking for how to bring our isolated tech tugboat back to the shiny cruise ship that was Cars.com’s replatformed universe.
I was fortunate to work alongside some really awesome folks over the years and means a lot. I’m so grateful for that. Because while I’m proud of the work and the cool things I created, it’s that I have a network of respect and trust, and well, validation too.
Since March I have been using the time to study, read, and build, because you know, AI is here to slay. (Till now, the joke was “JavaScript is here to slay”, but the new monster slayed the old monster.)
Artificial intelligence has actually been a recurring visitor throughout my career. I was drawn to neural networks, cognitive science, human-computer interaction, and usability. I found myself striving to understand how the brain works and how to replicate it, but also how humans work. What would it mean for the people using them, and what ethical responsibilities came with creating them. But running a 20 x 20 cortical layer with 400k connections, running the simulation was painfully slow at the time. My path took me elsewhere since.
But now it’s back! I’ve been exploring knowledge representation, graph structures, retrieval systems, software architecture, and developer tools. I’ve also been revisiting ideas that never quite fit the dominant machine learning narrative, including whether learning systems based on self-organizing cortical layers and circular feedback might eventually complement, or in some cases offer alternatives to, backpropagation. I’d love to see where it’s at now.
Looking forward now, I’m in a cockpit of a jetliner full of knobs and buttons and levers. I don’t which of them to twist or press or flip, or how much so that it will find me more fuel. I’ve got fuel in the tank but at cruising altitude, it’s definitely decreasing. I will find more fuel soon. Do I keep cranking one knob all day long hoping that one more turn will find the fuel (LinkedIn), or quickly go through every single instrument, barely waiting to see what happens before switching to the next one. I don’t know! Either way I could possibly spend all my fuel searching and never finding. This is the way I describe my multifaceted search for the right opportunity.
This post is one of those knobs. Here’s another one: https://darkbadge.info. Yesterday I wrote up a discussion of finding the missing link of hypertext: https://blog.shming.com/2026/07/24/beyond-hyperlinks-the-missing-layer-in-knowledge-navigation/
If we’ve worked together over the years, thank you. You had more influence on my career than you probably realize.
If you know of a team that might benefit from someone with deep engineering experience and a renewed interest in where computing is heading next, I would appreciate an introduction.
Leave a Reply