Product Manager - User & Context Intelligence

Glance runs an agent-first consumer platform across surfaces: app, lockscreen, Connected TV, browser, OEM integrations, ecommerce embeds, and partner journeys. Almost every personalization, recommendation, generation, and agent experience depends on a shared understanding of who the user is and what they want. However, when each surface builds that understanding on its own, signals fragment, quality drifts, and the same user looks like a different person on every surface.

We are looking for a Product Manager to own User & Context Intelligence as a reusable capability: the product direction for how user, visual, behavioral, and commerce signals become durable intelligence attributes and a taste graph that other teams all consume through clear contracts.

This role spans two tracks. The primary one is intelligence attributes and the taste graph: the reusable user-understanding products that other teams build on. The secondary one is signal capture and instrumentation: making sure surfaces emit the right events, at the right freshness and confidence, so those attributes stay trustworthy.

What you will own

Intelligence attributes and taste graph

• The attribute catalog: reusable user attributes across style and visual preference, occasion affinity, brand and merchant affinity, price sensitivity, commerce intent, and habit stage, exposed as versioned products with clear input, output, freshness, and confidence contracts.
• The taste graph built from purchase-proximal and engagement signals, served to personalization, recommendations, generation, and agent surfaces through a consistent interface rather than per-surface rebuilds.
• Coverage and confidence as first-class product concerns: attribute coverage across engagement-depth tiers, confidence scoring, and refresh cadence, so every consumer knows what a signal means and how far to trust it.
• The user model lifecycle from a product view: what gets modeled, how attributes are validated, how quality and drift are tracked, and how weak supervision extends coverage to sparse and cold-start users.

Signal capture and instrumentation

• The signal contract: what user, visual, behavioral, and commerce signals Glance captures across surfaces, and the events, entities, and attributes schema behind them.
• Instrumentation designed for reuse across reporting, personalization, model training, experimentation, and intelligence synthesis, not wired to a single consumer.
• Identity stitching, data quality, lineage, discoverability, and feedback loops, so attributes stay accurate as surfaces and behavior change.
• Demand-side clarity into Data Platform and ML Platform needs that matter to user intelligence: feature readiness, freshness, and serving requirements tied to real consumption from scientists, analysts, and surface teams.

What we are looking for

Must-haves, in priority order:

• User and preference modeling fluency. You understand how behavioral, visual, and commerce signals become durable user attributes and a taste graph, and how sparse, new, and cold-start users are handled without guessing.
• Signal and instrumentation design. You can define what to capture, at what granularity, freshness, and confidence, and design events for reuse across reporting, training, and personalization rather than for one surface.
• Platform product thinking. You think in contracts, versioned attributes, coverage and confidence guarantees, adoption metrics, and consumption, so intelligence ships as products instead of one-off pipelines.
• Applied science/ data science partnership. You work deeply with Applied Science on attribute models, weak supervision, validation, and evaluation, and can hold a technical position and negotiate with senior ML and engineering without formal authority.
• Adoption. You have a track record of getting surface and platform teams onto shared capabilities through workflow fit and clear contracts rather than mandates.
• Spec quality. You turn an ambiguous problem into a PRD an engineer or scientist can build from: scope in and out, owners, acceptance criteria, and edge cases.
• Judgment on the coverage-precision trade-off. You know when to extend an attribute to more users at lower confidence and when to hold the line, and you can defend the call.

Background

Strong product experience in one or more of: personalization, user modeling, customer data platforms, identity, recommendations, ML platforms, data platforms, or experimentation platforms. Direct experience owning a data-backed or model-backed product where attribute or signal quality was the core deliverable, ideally serving multiple internal consumers. Working understanding of feedback loops, cold start, weak supervision, and online and offline evaluation. Comfortable moving between technical and strategic registers, and crediting only what shipped.

This is a high-leverage role for someone who wants to build the user-understanding layer that agent-first experiences across many consumer surfaces depend on.