Staff Machine Learning Scientist

About the role

The role owns the machine learning that determines which message a member receives, when it is delivered, and through which communication channel. You will set the technical direction for send-time optimization, propensity modeling, and the experimentation rigor behind every nudge the team ships. As a Staff Machine Learning Scientist, you will write code that senior engineers respect and mentor a small machine learning team while partnering closely with product, data science, and growth and marketing colleagues. The ideal candidate has shipped recommendation or sequential-decisioning systems that changed real user behavior, runs experiments with rigor, and prioritizes what moves member outcomes rather than model sophistication for its own sake. You will own at least one model in production end-to-end, ensuring that systems perform reliably at scale. This position focuses on building systems that decide the right message for the right member at the right time, maximizing engagement and clinical impact. Your work will directly influence how members interact with their care plan and whether they stay on track with therapy.

Key facts

What you'll do

Design and ship the next system for deciding what nudge to send a member, when, and through which channel, moving beyond the current contextual-bandit approach.

Build and deploy models that decide whether nudging a given member is worth the risk of fatigue or unsubscribes, optimizing the tradeoff between engagement and retention.

Set the experimentation standards for the team, including multi-arm tests, sequential testing, CUPED, and safeguards against peeking to ensure causal validity of nudge decisions.

Own at least one model in production from data to deployment, handling monitoring, iteration, and performance under real-world conditions.

Mentor machine learning scientists on the team, guiding technical direction and elevating the quality of code and experiments.

Partner with product, data science, and growth and marketing teams to align model objectives with clinical and business goals.

Lead the implementation of recommendation and sequential-decisioning systems that influence member behavior at scale.

Establish best practices for evaluation and monitoring of models that influence user-facing communications in healthcare settings.

Drive the adoption of causal inference techniques beyond A/B testing, such as difference-in-differences and synthetic controls, to strengthen insights.

Explore low-data regimes and cold-start strategies specific to healthcare where per-member data can be sparse or noisy.

Contribute to infrastructure decisions involving Statsig, Databricks, feature stores, Airflow, and dbt within the existing technology stack.

Ensure that models comply with healthcare data constraints and privacy considerations, including familiarity with HIPAA and BAA requirements.

Collaborate with the growth and marketing teams to refine message targeting and channel selection based on empirical results.

Continuously iterate on deployed models based on monitoring data and experimentation outcomes to sustain long-term engagement.

Requirements

Hold a minimum of a Bachelor's degree or higher in Computer Science, Statistics, Operations Research, Machine Learning, or a related quantitative field.

Possess a minimum of 4+ years building and deploying ML systems in production at consumer scale.

Have shipped at least one recommendation, ranking, or sequential-decisioning system end-to-end, covering modeling, evaluation, deployment, monitoring, and iteration.

Demonstrate fluency in experimentation and A/B testing, including multi-arm tests, sequential testing, CUPED, and awareness of common failure modes in online experiments.

Show proficiency in Python and SQL, with the ability to read a colleague's pull request and improve it effectively.

Exhibit a deep understanding of machine learning and applied statistics, with a track record of making decisions that affect real users.

Have experience working with contextual bandits or reinforcement learning operated in production environments.

Understand multi-objective optimization balancing engagement, adherence, retention, and cost in real systems.

Have experience with causal inference beyond A/B testing, such as difference-in-differences, synthetic controls, or instrumental variables.

Comfortably work in low-data regimes and design modeling strategies for cold-start scenarios common in healthcare.

Have experience hiring and mentoring a small machine learning team.

Bring healthcare, fintech, or other regulated-data experience, including familiarity with HIPAA and BAA constraints.

Be familiar with the adjacent technology stack, including Statsig, Databricks, feature stores, Airflow, and dbt.

Possess familiarity with TypeScript as a beneficial additional skill.

Nice to have

Only include qualifications and experience explicitly mentioned in the source description.

Practical notes

The role is full-time based in San Francisco-HQ. No specific hours, travel requirements, visa information, or application deadlines are stated in the source.