About
I'm Nelson Koskela, a software engineer with four years of experience building production data infrastructure in Python, most recently at JPMorgan Chase & Co. I'm currently looking for the next place to put that experience to work, while also building out independent projects on the side. On this site you'll find a retrieval-augmented AI assistant I built from scratch, along with a live pipeline visualizer where you can submit a character and watch it move through an actual queued processing system. I also designed and built a dimensional data warehouse with its own automated test suite and self-directed adversarial review, applying that same modeling discipline to both data pipelines and AI systems.
Classical voice at Ruth Asawa
Before engineering, I trained in classical vocal performance at the Ruth Asawa School of the Arts in San Francisco, with a minor in jazz. I've always thought of myself as blending a technical and a creative background rather than choosing between them.
A Computer Science degree at Cogswell
I studied Computer Science at Cogswell University of Silicon Valley, after beginning my studies at Howard University in Washington, DC.
StreetCode Academy
Before JPMorgan, I was a Diversity & Inclusion Officer and freelance developer at StreetCode Academy, building responsive web applications with HTML, CSS, and JavaScript, and running sprints under Agile Scrum to deliver client solutions on time.
I also took on the organisation's own logistics and inventory tracking, building the analytics reporting that improved item-tracking accuracy by 50% — the same instinct for turning a manual, ad hoc process into something measured and reliable that later showed up in the regulatory work at JPMorgan.
Four years inside a bank's engineering org
I joined JPMorgan Chase & Co. in July 2022 after interning the previous summer through the Software Engineering Program, and spent the next four years as a Software Engineer II in the Corporate & Investment Banking technology organisation, engineering and operating data pipelines and analytics infrastructure for regulatory and risk reporting — reliably processing multi-terabyte datasets across global business units in a production, always-on environment, and partnering directly with business stakeholders to deliver that data in live, end-of-day, and start-of-day formats.
I later stepped up as de facto project lead on a full system migration during a team staffing shortage, and separately designed and led a parallel “prod vs. prod” validation framework end to end — automated reconciliation that guaranteed two independently-built platforms matched to the record before every release.
I'm Nelson Koskela, a software engineer with four years of experience building production data infrastructure in Python, most recently as a Software Engineer II at JPMorgan Chase & Co. on the Corporate & Investment Banking side. I joined the bank in July 2022 after interning the previous summer through their Software Engineering Program. These days I'm looking for the next place to put that experience to work, while also building out a handful of independent projects on the side.
My path here wasn't a straight line. I started at Howard University in Washington, DC, then finished my Computer Science degree at Cogswell University of Silicon Valley. Before any of that, I trained in classical voice at the Ruth Asawa School of the Arts, with a minor in jazz. I'm still the kind of person who's curious about a lot of things at once, and I care about building technology that actually improves someone's day, not just technology for its own sake.
On the technical side, I work mostly in Python and production data systems, and I've put real time into building and shipping agentic AI applications, including a retrieval-augmented AI assistant that runs on this site. I also designed and built a live pipeline visualizer here, a small simulated world where you submit a character and watch it move through an actual queued processing system, so if you have a minute, send one through. Separately, I designed and built a dimensional data warehouse from scratch, complete with an automated test suite and a self-directed adversarial review, applying that same modeling discipline to both data pipelines and AI systems.