Data Scientist

Stealth Startup; New York (NY); On-site; USD 200,000 - 225,000; Full time

Benefits offered by this job

Relocation package

Competitive equity

Marquee benefits package

Job summary

Stealth Startup is seeking a Data Scientist who builds features and models powering a consumer graph and agents atop it. You own the path from raw data to production-ready features that agents can call at scale, shaping the foundation for other systems as we grow.

You’ll expand predictive coverage, improve income/wealth models, and contribute to a robust data lake while deploying pipelines and validating outputs for autonomous use.

Qualifications

• 2-4 years working as a data scientist, data engineer, or software engineer in a data-heavy context.
• Proficient in Python and SQL with a data-driven mindset.
• Reason from first principles and translate datasets into usable features.
• Strong data engineering intuition around cleaning, ingestion and modeling.
• Solid engineering background with production pipeline deployment experience.
• Willingness to work in the NYC office; relocation package provided.
• Flexible and eager to learn and grow with the team.

Responsibilities

• Create new features for models and agents and expand the predictive surface.
• Improve existing models with rigorous feature engineering on income/wealth signals.
• Contribute to building out the data lake and making terabytes of data queryable.
• Develop feature pipelines that operate reliably at terabyte scale, production-ready.
• Write validations and guardrails so agents can act autonomously.

Skills

Python

SQL

Data engineering

ML engineering

Tools

Dagster

dbt-core

Spark

Iceberg

Trino

AWS Glue

Modal

Job description

About the Role

As a Data Scientist you build the models and features that power our consumer graph and the agents that run on top of it. You sit at the intersection of heavy data engineering and applied modeling: you architect feature engineering pipelines that are computed over terabytes of data, train and sharpen the models that drive targeting and prediction, and ensure the outputs are robust enough to be consumed autonomously by our Minerva Agents and our world-class modeled attributes (i.e. income / wealth).

This is a role that will be deploying constantly to production. The models you build are not handed off to be deployed by someone else, you own the path from raw data to a feature or model that an agent can call reliably at scale. As we grow, your work becomes the foundation other systems are built on.

What You’ll Do

• Create new features for models and agents, expanding the predictive surface area of our consumer data lake and building the pipelines that turn raw signal into trusted attributes.
• Improve existing models through rigorous feature engineering, including our income/wealth, home buyer, and home seller models.
• Play a pivotal role in the buildout of our world-class data lake, shaping how terabytes of consumer data are stored, transformed, and made queryable for both humans and agents.
• Build feature engineering pipelines that run efficiently at terabyte scale, with the data engineering rigor to make them reliable in production. This is a 70/30 split DS/DE role.
• Ensure model and feature outputs are reliable enough to be consumed agentically, writing the validations and guardrails that let our agents act on your work without a human in the loop.

Our Data Stack

• Dagster for all things orchestration
• dbt-core within Dagster as the primary data transformation surface
• Spark, Iceberg, Trino, AWS Glue for Lakehouse workloads
• Modal for ML eng
• Frontier OSS models & agent SDKs. We are heavy users of OpenAI/Anthropic batch APIs

Qualifications

• 2-4 years working as a data scientist, applied machine learning focused data engineer or software engineer in a data-heavy context. Simply put, you live and breathe data.
• Highly proficient at Python and SQL.
• You are driven by first-principles thinking and are a go-getter. You reason about what datasets and features are necessary to solve a modeling problem, and are scrappy and clever enough to bring that to life.
• Strong intuition for data engineering principles, especially around data cleaning/ingestion and data modeling. We prefer these core skills to be second-nature, freeing up thinking for architecting and executing large-scale data initiatives, especially given the advancement of AI coding tools.
• Strong engineering background. You are comfortable deploying complicated production pipelines and working within larger production systems, not just in sandboxed or research environments.
• Willingness to work in office in NYC (we provide a relocation package).
• Flexibility and openness to wearing several hats. We are lean and things are always changing.
• Eagerness to learn and grow with the company and your coworkers.

Preferred

• Experience building and training predictive models (e.g. lead scoring, LTV, propensity, lookalike modeling).
• Experience with orchestration tools like Dagster, Airflow, Prefect and SQL transformation tools like dbt, SQLMesh.
• Experience with both transactional databases (e.g. Postgres, MySQL) and analytical databases (e.g. Snowflake, Redshift), with a bias toward the latter.
• Familiarity with a cloud resource provider (e.g. AWS, GCP).
• Familiarity with backend and ML/AI engineering.
• Experience with AI coding tools (e.g. Cursor, Claude Code, OpenCode) as a force multiplier.
• Prior work at an early-stage startup.

You don’t need to tick every box. If you’re strong on the engineering side and hungry to build models that matter, we want to hear from you.

Compensation

Base salary: $200,000 to $225,000, commensurate with experience. Competitive equity and a marquee benefits package.