About us
Complir is the AI platform for the product compliance lifecycle, built for retailers, brands, and manufacturers selling physical consumer products across multiple markets.
We believe getting a physical product to market shouldn't mean fighting spreadsheets, PDFs, and supplier emails just to prove it's safe and legal to sell. Product compliance should be continuous, automatic, and understandable, not a bottleneck that slows launches down. We're building the AI infrastructure that makes that possible.
We work with many of the world's largest retailers and brands across fashion, furniture, cosmetics, toys, electronics and many other product categories. We're backed by Y Combinator, and a roster of investors who see what we see: Product compliance is the next major infrastructure category, and we're leading it.
About the role
Every product you've ever bought, from a kid's toy dinosaur to your electric toothbrush, had to clear a wall of regulation before it reached you. That wall is written down somewhere: scattered across hundreds of government portals, PDFs, and legal databases, in dozens of languages, updated on no schedule anyone agreed to.
Your job is to turn that chaos into structured, queryable data our product can act on. You'll be the person who makes the world's product regulations machine-readable.
The data layer everything stands on
This is not a role where you write Python scripts in isolation. You'll understand, crawl, and monitor regulatory sources across hundreds of jurisdictions, parse raw legal text into a normalized requirement schema, and catch changes the day they land, so every label, Declaration of Conformity, and risk assessment we generate is provably current.
The interesting problems are everywhere: government portals that were last redesigned when fax machines were current, legal text that means different things in different translations, amendments that quietly rewrite three other acts, and the question of how you prove to an enterprise customer that your data was correct on the day they shipped. LLMs make structured extraction over legal text possible for the first time; making it reliable enough to stake a product launch on is the engineering problem you'll own.
You'll do this sitting next to the founders, our AI engineers, and our in-house compliance experts at our Copenhagen HQ. No layers, no ticket queues. When a source breaks or a regulation changes, the fix ships in hours, not sprints.
What you'll actually do
Build and operate the crawling and monitoring layer across hundreds of regulatory sources: government portals, legislative databases, and official gazettes in dozens of languages
Parse raw legal text (HTML, PDFs, scans) into a normalized requirement schema, using LLM-assisted extraction pipelines you design, evaluate, and harden
Build change detection that catches new acts, amendments, and repeals the day they land, and triggers the downstream updates our customers depend on
Own data quality end to end: provenance, versioning, and coverage metrics, so every requirement in the platform traces back to its primary source
Work with our regulatory experts to evolve the schema and taxonomy as coverage grows from the EU to new jurisdictions
Shape the architecture, tooling, and data culture as the team grows
You will love this job, if you are:
An engineer with several years building data pipelines, crawling infrastructure, or large-scale extraction systems, fluent in Python
Experienced with (or seriously hungry to go deep on) LLM-assisted extraction: structured outputs, evals, and the failure modes of models reading dense text
Someone who treats data correctness as sacred: provenance, versioning, and "how do we know this is still true?" are your favorite questions
Unafraid of messy sources: broken HTML, scanned PDFs, and legal prose in languages you don't speak are puzzles, not blockers
Product-minded: You understand that a quality manager will stake a product launch on your pipeline's output
Bonus points if you have worked with regulatory data before
Comfortable with ambiguity and speed: Small team, fast decisions, no spec handed down from above
Excited to work on-site in Copenhagen with the whole team
A builder at heart
What you get
Data engineering with actual stakes: Your pipelines decide whether "can I legally sell this product?" gets answered with current law or last year's
Ownership of the regulatory data layer, the foundation the entire product stands on
Zero layers: You sit next to the founders, the AI team, and the compliance experts whose domain you're turning into data
Ground floor of a category-defining company, backed by Y Combinator and investors betting on product compliance becoming the next big thing
Competitive salary and equity
An office where the pastries are mandatory
If you've read this far and thought "that's the data problem I want to own," we'd like to meet you.