Senior Product Manager, Recommendations

About the role

Ali, our founder and CEO, sets company priorities. You own the ideas. You will spend part of every day researching competitors, the industry and our users to generate new feature ideas, and you will keep a live Top 3 list of the highest-impact ideas for your area, a new idea only enters by beating one already on it.

What you'll do

• Own recommendations and discovery across Inkitt, Galatea, CandyJar and Ironblood, every surface where we decide what to put in front of someone next, plus the roadmap, experiments and revenue attached
• Design the experiences the models feed: home and feeds, search, browse, taste capture at onboarding, personalised notifications and re-engagement, cross-app discovery
• Partner closely with Data Science, who own the models and signals, on what those models need to predict and how we judge whether they got better
• Design and run recommendation experiments at a high cadence; ramp them on your own judgement; report results honestly including the losses
• Build the impact case before you build the feature. Know the difference between an offline metric moving and revenue moving
• Hit the impact you predicted

What they're looking for

• 5, 12 years of professional experience. We measure years on the job, not years holding a “product manager” title. If you came to product from data science, analytics or engineering, that is an advantage here rather than...
• You have shipped a recommendation, ranking, search or personalisation system that real users felt, and you can explain what you changed, what moved, and how you knew it was causal.
• You understand how these systems work underneath. Candidate generation and ranking, cold start, feedback loops, popularity bias, exploration versus exploitation. You can hold your own with the people training the models.
• You can tell an offline win from a business win. Plenty of people have improved an NDCG. We care that you have moved reading time, conversion or renewal, and that you can show the experiment design behind it.
• Numbers you produce yourself. Advanced SQL. You build the analysis rather than commissioning it.
• Consumer product instinct, this is a reader-facing experience, not an internal tool.
• Based in San Francisco. This role is five days a week onsite.
• Nice to have: content or entertainment recommendations (streaming, serialized fiction, short-form video, audio, games); LLM-based ranking, embeddings or generative approaches to catalogue understanding; experience with...

About CandyJar (Inkitt)

CandyJar (Inkitt)

All jobs →