Unzippd: Building An Inclusive Data Platform
How a data layer that makes answers of deaf users readable hired Full-Stack Engineer and Data Architect.

Challenge
Unzippd is building the UK’s first two-way BSL research platform. British Sign Language is a language in its own right, with its own grammar, and for the 22,000 people in England and Wales who record it as their main language, written English is a second language rather than a neutral medium. A text survey is therefore not a smaller version of the same research instrument. It is a different instrument, one that quietly filters out the people it is meant to reach. Unzippd’s answer is to ask questions in BSL and collect answers in BSL, which turns every question and every response into video.
The request was for a Full-Stack Engineer and a Data Architect, with no in-house engineering team. The platform needed a front end that treats video as its primary input rather than as an attachment, and that performs on the devices and connections the audience actually uses: React, Next.js, TypeScript and Angular, with state management architecture and automated testing behind it. The second gap was harder to see from the outside. Video responses are not survey data, so someone had to design the pipeline that turns thousands of recordings into something readable: Python, Spark, Airflow, Snowflake, Terraform and AWS, with IAM and access control modelled from the start.
Profiles for a Full-Stack Engineer request were engaged within 48 hours, followed by a Data Architect. We matched against whether the engineer had built interfaces where performance was a product requirement rather than a preference, and whether they had designed governance and access control for regulated data rather than inherited someone else’s. Founder interviewed and chose both engineers themselves, and the engagement has been running for over six months.
Outcome
Two engineers - one that owns what the Deaf user sees and records, and the other owns what happens to those recordings afterwards, both matched within 48 hours.
Both roles matched within 48 hours of the requests opening, with no internal recruiter and no separate agency per hire.
- Both roles matched within 48 hours of the requests opening, including the complex and compliance-heavy Data Architect one.
- A complete platform team of two: a video-first interface, and the data architecture underneath it.
- Both roles are filled with no technical leadership or engineering management.
- Over six months in, engineers are still on the engagement delivering new features and meeting data requirements.
Now on second project with Yotewo. Responsive team who bend over backwards to help and accommodate requests. Would definitely recommend collaborating with them. Platform is easy to use and navigate.

A Strategic, Multi-Phase Hiring Approach
Phase: Request
Two consequential requests were opened and defined by stack and by the problem each role owned: React, Next.js, TypeScript and Angular for the Full-stack Engineer, and Python, Spark, Airflow, Snowflake, Terraform and AWS for the Data Architect.
Phase: Matching
6 engineers were shortlisted across the two requests and 4 went to interview, each one vetted before founder of Unzippd received their profiles.
Phase: Interviews
Interviews were booked within hours of the request opening, and both engagements were confirmed within 48 hours.
Phase: Engagement
Both engagements opened with a part-time monthly workload, an hour cap set beforehand, daily hour commitments in the tracker, and monthly reports approved over 6 months.
Phase: Support
Across six months the two engagements have run with workload and hour cap changes requested, approved and recorded.
