Marco Altran
Senior Software Engineer.
Large-scale distributed systems where the work has a sharp connection to the product. Crawl and index pipelines handling over 100 million pages a month with Kafka and Vespa. Customer support LLM agents over millions of help docs. A regulated digital bank, top to bottom. The common thread is systems other people use every day, where reliability and observability matter.
Taking systems from "kind of works" to "ships on data, not vibes." Most recent example: bumping an AI support agent's resolution rate from 10% to 70%, which translated to about $500k a year in deflected revenue. The model wasn't the hard part. The eval pipeline was.
A regulated digital bank, from scratch, in 12 months. I was the first employee: picked the stack, hired the team, designed the architecture, dealt with regulators. We hit 10,000 accounts in the first three months.
Both, well. Currently IC. I've led teams of nine and I've also been heads-down on a thorny indexing pipeline. I gravitate toward whatever the work needs.
Anything where the engineering has a sharp connection to the product. Search, applied AI, the infrastructure that makes either of those work. I'd rather be solving real problems for real users than polishing abstractions.
I was a professional poker player from 2009 to 2014, with $100k+ in profits. The same instincts that worked over thousands of hands turn out to apply to system design and incident response. Read the room, take the right risks, leave the ego at the door.