Ads AI Analytics Lead II

Instacart Instacart · Consumer · Canada · Remote · Commercial Excellence

Instacart is seeking an Ads AI Analytics Lead II to own the intelligence behind their Ads agents. This role involves designing the Ads semantic/context layer, building vertical AI agents to analyze campaigns, diagnose performance, and recommend actions to improve ROAS, pacing, and partner outcomes. The position requires partnering with GTM, Product, Data Science, and Engineering to ship production agents with measurable lift. Responsibilities include defining Ads ontologies and metrics, building dbt models, ingesting and enriching unstructured Ads content, designing and evaluating retrieval workflows (RAG), designing agent reasoning and policies, establishing evaluation suites, running A/B experiments, and translating Ads problems into agent behaviors.

What you'd actually do

  1. Define Ads ontologies and metrics for campaigns, budgets, bids, creatives, audiences, and placements.
  2. Build dbt models and curated marts in Snowflake with clear data contracts, tests, and SLOs.
  3. Ingest and enrich unstructured Ads content and publish retrieval‑ready datasets using our managed search/vector services.
  4. Design and evaluate retrieval workflows (RAG) with existing services for hybrid search and re‑ranking; set quality/latency targets and iterate via experiments.
  5. Design agent reasoning and policies on ads, including tool definitions and human‑in‑the‑loop approvals.

Skills

Required

  • SQL
  • Python
  • dbt
  • Snowflake
  • data modeling
  • data contracts
  • data visualization tool
  • offline/online evaluations
  • A/B or uplift tests
  • Ads analytics concepts
  • stakeholder communication
  • ML models
  • evaluation frameworks
  • guardrail frameworks
  • human‑in‑the‑loop QA

Nice to have

  • Airflow
  • BigQuery
  • forecasting
  • anomaly detection
  • uplift modeling
  • causal inference
  • workflow automation
  • low-code development platforms
  • retail media or ad platforms

What the JD emphasized

  • shipping production data or AI systems that drove business impact
  • evaluation and guardrail frameworks
  • human‑in‑the‑loop QA

Other signals

  • design and build vertical AI agents
  • analyze campaigns, diagnose performance, and recommend actions
  • ship production agents with measurable lift
  • define Ads ontologies and metrics
  • design agent reasoning and policies
  • establish evaluation suites
  • run A/B or uplift experiments
  • translate Ads problems into agent behaviors