Senior Data Scientist

We have partnered with a consumer entertainment platform now working on a very cool consumer entertainment platform. This is a small funded team with a runway for the next 3 years and they will be launching their product in December this year. Users of this app will be able to get personalised feeds of what they follow across games, sports, music, TV, film, podcasts and more and there are agents in the app also working for you to help you find exactly what you are looking for.

Take the next step in your career now, scroll down to read the full role description and make your application.

In this role you will own the recommendation and search models behind the product. The platform is already built and running: a Databricks lakehouse, a Neo4j property graph, vector search, and live ranking and search endpoints serving a real corpus. Infrastructure, CI/CD, ingestion and content sourcing are owned by others. You will write models in this position and not the plumbing.

Near term is extending the existing recommender to new content categories ahead of launch and getting the personalized feeds to their best possible output. Post-launch the role opens into a large knowledge graph problem with genuine architectural ownership.

This role will suit someone who has built and shipped recommendation or search ranking systems that real users interacted with, at meaningful scale. Typical backgrounds that fit:

• Product or search ranking at an ecommerce marketplace, deciding what to surface on a home page, category page or "similar items" module

• Content or feed ranking at a media, music or streaming company

• Candidate retrieval and ranking at an ads or jobs platform

• Personalization at a subscription consumer app, including cold start for new users and new items

• Retrieval engineering on a hybrid vector plus knowledge graph system, in any domain

The common thread is production systems with measured outcomes, not research or offline prototypes.

Must have

• Production recommender or ranking systems, shipped to real users

• Hybrid retrieval: combining vector similarity with graph or structured signals, and deciding how the results are weighted and blended

• Databricks and Python, current and hands-on

• Graph experience. Neo4j and graph data science strongly preferred. This is the differentiator between candidates. xhyhwjd

• Comfortable becoming productive fast in a substantial codebase they did not write

Nice to have

• Offline evaluation, A/B testing frameworks, ranking regression testing

• Embedding and prompt design for retrieval quality

• Cold start on sparse or newly launched content categories

If this role is of interest, please apply here and also send a copy of your CV to and I will review and get back to you as quickly as possible.