Senior Recommendation Engineer

NewsBreak•Mountain View, CA

About The Position

NewsBreak is seeking a Senior Recommendation Engineer to build and own recommendation and matching systems for their new company, Nearby AI. This role involves developing systems that go beyond optimizing for the next click, focusing instead on supporting infrequent, expensive, high-consequence household decisions where the right recommendation might be to do nothing. The engineer will build recommendation and matching across three surfaces: content feed personalization, an event-triggered engagement system, and two-sided matching between customers and service professionals with fairness constraints. Ranking will not be for sale, and objective functions must reflect this principle.

Requirements

5+ years of machine-learning or software engineering, including 3+ years shipping recommendation or ranking systems in production.
Built and operated a ranking system at consumer scale, millions of users, including candidate generation, ranking, and monitoring.
Substantial online experimentation experience on ranking changes, and can explain at least one test that failed and why.
Deployed an uplift, causal, or counterfactual model in production, or can give a rigorous account of why click-based objectives are wrong for rare, high-cost decisions.
Strong Python plus at least one of Java, Scala, or Go; production experience with feature stores and streaming pipelines.
Ability to explain modeling decisions to product and business stakeholders and to defend experimental design under commercial pressure.

Nice To Haves

Feed or notification ranking at a content or media company.
Two-sided marketplace matching with fairness or exposure constraints in production.
Domains with low-frequency, high-consequence decisions such as insurance, healthcare, or real estate.
Publications or open-source contributions in recommendation, causal inference, or marketplace design.

Responsibilities

Build and own the recommendation and ranking systems for feed surfaces, from candidate generation through online experimentation and monitoring.
Design the event-triggered engagement engine with the messaging platform team: condition detection, propensity and uplift modeling, frequency management, and holdouts.
Build customer-to-professional matching, starting with transparent rules and evolving to learned ranking, with an explicit fairness and exposure contract so new, high-quality providers can win work.
Define evaluation frameworks that treat downstream outcomes and negative signals, including complaints and "do not proceed" recommendations, as first-class objectives.
Contribute learnings back to the company's pricing and intent models owned by the AI team.