Hey, I am Jimmy 👋
I’ve been a software engineer for over two decades, touched a variety of tech stacks, domains and whatnot—some might say too many, I’d say that I had a blast learning about each domain, and most skills are transferable.
I would describe myself as a backend “full-stack” engineer if there’s such a thing. Platform engineering at the core, real-time applications are my preference. But I’ve also worked on data ingestion, data pipelines, various aspects of ML, feature engineering, data cleanup, and a hint of infra. I did build reactive UI, so this might actually make me full stack, but I wouldn’t put my money on my frontend skills :p
Shipping a product that is used is what gets my juices going. I enjoy tinkering with tech, sure, but I prefer shipping products that solve problems and scale. Having been on-call for what I ship, I value shipping something useful, robust, and reliable more than the tech it’s built with.
Observability is one of my obsessions—very visual, I love seeing dashboards that tell me everything works and cannot stand having to read logs. By far, I’d rather trust tests and metrics than my manual testing skills. Some of the best apps I’ve worked on are ones where I confidently ship changes without ever feeling I need to run the app.
I do enjoy the incremental approach. There’s always a mid- to long-term target and more often than not, a POC to showcase value. It’s fine if we don’t get it right the first time as long as we learn from our mistakes. I haven’t seen a lot of “stop the world and rewrite the entire stack” situations in my career, so continuous improvement is what I think will almost always get the job done.
What have I been up to?
Data Movement
I’ve been around this space for a while—sometimes as a data-consumer trying to figure out how to get my data, sometimes as a data producer, sometimes both. Most recently, I’ve been a data mover: solving this hard problem for others by building and scaling Airbyte, an open source data movement platform.
Airbyte has many different dimensions. The two I know best are dealing with the sheer amount of data movement that needs to run (platform/infra side) and making each of those jobs fast and reliable.
Working on a Cloud/OSS platform is an odd one. Having spent a lot of time in SaaS, I know migrations are generally challenging. But nothing compares to OSS. I remember thinking migrations were painful—that’s nothing compared to ensuring a smooth transition for open source users who deploy and upgrade on their own cadence.
I’ve also worked on real-time ingestion data pipelines at various scales, though I’ve probably spent more time on the producer or consumer side than maintaining the pipes themselves.
Sports Tech
I do enjoy sports and sport science. Working on a sports wearable was great. There’s always something more exciting about hardware, the tactile feel and constraints that come with the three dimensional space.
As a software engineer, this gave me exposure to the full dev cycle of garments, hardware, and firmware. Most compile-test loops don’t sound too bad once you’ve seen a test cycle involving 3D printing.
As a side benefit, I left with knowledge on how to structure workouts and minimize injuries. Not a bad takeaway.
AdTech
I never thought I’d be working on ads, but beyond funding the web, it actually has a certain appeal once you get past the obvious. Data at scale, lots of it, high-performance computing where milliseconds matter—clickthrough rates decrease with time to display, for example.
I ended up touching most of the pipeline. It started with user tracking—the challenge of JavaScript and pixel services that have to respond in milliseconds, or you hurt advertisers’ and publishers’ page load time. I thought cache lookups were cheap back then. That’s not true when you’re running in the sub-100ms range or you lose your opportunity.
Then I moved into data pipelines for reporting and ML. Making sure jobs scale without hotspots, ensuring you don’t drop or mutate data unexpectedly. The latter is the fun one because there are always nasty debugging sessions for ML consumers when a model was trained on an int but evaluated as a string.
Feature engineering and recommendation systems came next, because why not. It’s where intuition and data-driven insight play together—or don’t. I found working with folks in this space is always interesting. You can be entirely data-driven, but intuition can make you move so much faster. At the same time, intuition without solid data validation can backfire hard.
Text Mining
I started my career as a NLP engineer. This might sound dated, but I was very involved in the previous generation of unstructured data processing. Ngrams, tokenizers, taggers, and named entity extraction were my bread and butter. I built compilers that would generate named entity extractors, sentiment analysis, and text generation systems.
Diving into LLMs and friends, my first impression: the fundamentals haven’t changed. LLMs still run on those concepts. The need to optimize each step evolved, but understanding those underlying layers remains valuable.
Tech Stack
As a student, I somewhat thought that I’d be Unix/C++ for life. Little did I know that I would be shipping production code in Python, C#, Java, and Kotlin as well. C# on Windows was unexpected for sure, turns out that Visual Studio might be one of my favorite IDEs, close second to nvim.
I’m not the infra guy, but I’ve deployed code everywhere—bare metal data centers and clouds across GCP, AWS, and Azure, from old-school installs to Kubernetes. Cloud is convenient, but bare metal data center definitely had its charms. Sharks apparently have a sweet tooth for fiber cable, and fire in a cage potentially leads to massive fail-over and restart that takes down the DB before everybody’s up and running.
And of course, a beloved Raspberry Pi Kubernetes cluster that’s going to enjoy those lightweight Go applications if I can figure out how to make Claude write some Go for me -_-'
What Else?
I think I missed the courses about Resting. As a result, my hobbies are all over the place.
Snowboarding in the winter, tree runs are fantastic, and there’s always the initial thrill of staring down the mountain before dropping in. I never thought I was a speed demon but I do enjoy racing and sim-racing 😈
Bouldering and calisthenics got to me. I used to be more of the interval training and plyometrics kind. I guess solving problems and working hard towards an unattainable planche just work for me.
I love playing guitar and drums, maybe one day I’ll figure out how to properly work a keyboard. Always with some music to listen to, from distorted riffs to chiptunes with a hint of whatever’s in between. And… probably a whole lot of distortion, somehow this last one really doesn’t want to leave me alone 🤘
If you’ve made it this far and still want to chat more, find me on LinkedIn or GitHub.