Recommendations Product Jobs

1 job found

Today

Inkitt San Francisco, CA

San Francisco, CA

Today

$170k - $210k/yr Full time Data Analytics Product Analytics Product Management

Lead end-to-end product for recommendations and discovery across Inkitt’s apps, owning the roadmap, experimentation, and revenue impact.

Own recommendations and discovery across Inkitt, Galatea, CandyJar, and Ironblood, including the roadmap, experiments, and revenue tied to every surface that determines what users see next
Design model-fed user experiences including home and feeds, search, browse, taste capture at onboarding, personalised notifications and re-engagement, and cross-app discovery
Partner with Data Science on what models and signals must predict and on how success is measured when models improve
Run recommendation experiments at high cadence, using your own judgement to ramp tests and reporting results honestly, including losses
Build impact cases before building and differentiate between offline metric movement and business outcomes
Deliver to predicted impact

5–12 years of professional experience (years on the job, not years with a “product manager” title); transitioning from data science, analytics, or engineering is an advantage
Shipped a user-facing system (recommendation, ranking, search, or personalisation) that real users felt, with clear examples of what changed, what moved, and why you can say it was causal
Strong understanding of recommendation mechanics, including candidate generation and ranking, cold start, feedback loops, popularity bias, and exploration versus exploitation; able to work effectively with teams training models
Can connect offline wins to business outcomes, prioritizing changes in reading time, conversion, or renewal, plus the experiment design behind the result
Numbers and analytics done personally using advanced SQL to build analysis rather than commissioning it
Consumer product instinct for a reader-facing experience, not an internal tool
Based in San Francisco for five days a week onsite

401k plan
Health benefits tailored to your needs: medical, dental, and vision coverage
Professional coaching
Team-building events, including an annual Tulum trip
Unlimited access to the Galatea app and CandyJarTV app
Unlimited budget for self-development books
Charity donation of your choice at your one year anniversary
Free lunch in office every day
Class Pass membership for US-based employees and gym access for Berlin employees
Dog-friendly offices in Berlin and San Francisco

Own recommendations and discovery across Inkitt, Galatea, CandyJar, and Ironblood, including the roadmap, experiments, and revenue attached to what users see next
Design experiences the models feed: home and feeds, search, browse, taste capture at onboarding, personalised notifications and re-engagement, and cross-app discovery
Collaborate with Data Science on model prediction targets and evaluation of model improvement
Design and run recommendation experiments at high cadence, ramping based on your judgement and reporting outcomes honestly, including losses
Create impact cases before building and distinguish offline metric changes from revenue movement
Achieve the impact you predicted

5–12 years of professional experience; years measured on the job
Proven ability to ship recommendation/ranking/search/personalisation systems with causal learnings tied to real user impact
Working knowledge of candidate generation and ranking, cold start, feedback loops, popularity bias, and exploration versus exploitation
Capability to assess offline versus business impact, especially reading time, conversion, or renewal, backed by experiment design
Advanced SQL skills and the ability to build the analysis directly
Consumer product instinct for reader-facing outcomes
San Francisco based; five days a week onsite

Experience with content or entertainment recommendations (streaming, serialized fiction, short-form video, audio, games)
Experience with LLM-based ranking, embeddings, or generative approaches to catalogue understanding
Experience with a catalogue created in-house rather than uploaded by users

Ali (founder and CEO) sets company priorities
You own the ideas
Spend part of each day researching competitors, the industry, and users to generate feature ideas
Maintain a live Top 3 list of highest-impact ideas for your area; a new idea enters only by beating one already on the list
Pitch using a single slide with a rough impact calculation attached
Ideas that will not move the number enough are rejected, leading to an iterative pitch, sharpen, and re-pitch loop alongside delivery
Once approved, you own execution end-to-end, including release and A/B ramping
Four weeks after launch, features are reviewed against the impact predicted before building
Performance is measured on how often you beat your own Top 3 and on realised impact of what you ship

New Recommendations Product Jobs, daily.