Description

You will work on end-to-end problem framing, build models using modern ML and deep learning techniques, design sub-agents for reasoning and action, and ship production systems serving millions of users. Working primarily with Product and Engineering teams, you will leverage petabyte-scale execution data spanning two decades to translate complex modeling behavior into clear recommendations for stakeholders.

Responsibilities

• Design and ship AI sub-agents that act across the customer lifecycle, combining predictive models, retrieved context, and LLM reasoning to recommend or take action
• Build predictive and prescriptive models that power sub-agents, addressing churn risk, growth, adoption trajectories, account health scoring, and similar lifecycle problems
• Develop data foundations and knowledge layers that sub-agents reason over, applying responsible aggregation and privacy-aware design
• Design tools, retrieval, and grounding strategies each sub-agent uses; determine when a sub-agent should act, recommend, defer, or escalate
• Build evaluation harnesses that determine sub-agent production readiness and catch regressions in production
• Define metrics and experimentation strategy for sub-agent rollouts; measure real customer impact beyond offline accuracy or evaluation scores
• Partner with Product, Engineering, and Applied AI teams from problem framing through production deployment
• Drive a data and modeling culture within Product and Engineering, and mentor other data scientists on the team

Qualifications

• Required
  • Bachelor's degree and 8+ years of experience (or 10+ years of experience without an advanced degree)
  • Deep applied ML expertise across traditional ML and deep learning: gradient boosting, regularized linear models, transformer-based sequence models, foundation model embeddings, causal ML, contextual bandits, and offline reinforcement learning
  • Strong grasp of causal inference for intervention design and lifecycle modeling: uplift modeling, difference-in-differences, propensity scoring, and synthetic control
  • Solid foundation in statistics and experimental design: hypothesis testing, power analysis, multiple comparisons, sequential testing, and quasi-experimental methods
  • Hands-on experience taking LLM- and agent-based systems to production: tool use, retrieval, multi-step reasoning, evaluation, and guardrails
  • Experience operating ML in production — feature engineering and pipelines, model monitoring, drift detection, retraining cadence, and trade-offs between batch and real-time serving
  • Proficiency in SQL and Python
  • Comfort with ML/LLM tooling at scale (Spark, Databricks, Snowflake, or equivalents)
  • Experience with ML frameworks (PyTorch, scikit-learn, XGBoost/LightGBM)
  • Experience with visualization tools (Tableau or similar)
  • Experience modeling the customer lifecycle — churn, expansion, adoption, plan health, lead/account scoring
  • Business fluency in SaaS metrics that drive growth: NRR, GRR, ARR, and cohort economics
  • Pragmatic production bar: understanding of latency, cost, monitoring, drift, hallucination, and failure modes
  • Strong track record of forming effective cross-functional partnerships and communicating analysis clearly to technical and executive audiences
• Preferred
  • Advanced degree in a quantitative field (Statistics, Computer Science, Machine Learning, Economics, Operations Research, or similar)
  • Ability to research and learn new technologies, tools, and methodologies, and to thrive in dynamic environments

Key Skills

Education

Any Graduate