Backend Engineer, Data

Minerva•New York City, NY

1d•Onsite

About The Position

Minerva is hiring an individual-contributor or senior individual-contributor product engineer to make first-party consumer data and data-science models usable in its marketing product. The role works day to day with data engineering and data science, with the Staff Data Engineer as the warehouse-side counterpart. It owns the product layer between the company’s Snowflake data platform and marketers’ workflows, including performant first-party fields, model scores, dynamic segments, and an in-app agent that translates questions into queries against a brand’s data. The engineer makes deliberate choices about what to precompute, where it is stored, how it stays current, and how data moves from Postgres to the synced client. Most work builds on existing data-team output, but the engineer also writes models or pipeline steps when needed. The role ships end-to-end features across Python AWS Lambda services, Zero schemas, and TanStack Start pages. The stack includes TypeScript, TanStack Start, React 19, Tailwind, Base UI, Vite, Bun, Zero, Postgres, Drizzle, Mastra, Temporal, Python, AWS Lambda, Snowflake, semantic views, and Cortex. The position is based in person at Minerva’s office at 300 Kent Avenue in Williamsburg, Brooklyn; candidates should be in New York City or willing to relocate. In the first 90 days, new scores and first-party tables should reach the app through a repeatable system built by the engineer. Within six months, any first-party field should be sortable, filterable, and chartable without a live warehouse query, and the in-app agent’s query tool should run on the engineer’s layer. Applicants must include a demo of relevant work, such as a link, repository, or two-minute video; applications without a demo will not be reviewed. Candidates must be authorized to work in the United States. Visa sponsorship is considered case by case.

Requirements

Production experience with a cloud data warehouse; Snowflake is preferred.
Production TypeScript experience.
Experience shipping backend services on AWS, such as Lambda or similar.
Current knowledge of AI tooling and changes in the latest releases of packages used.
Personal projects or other things built outside work that can be demonstrated; the application requires a relevant demo.
Located in New York City or willing to relocate.
Authorization to work in the United States; visa sponsorship is considered case by case.

Nice To Haves

Experience with Snowflake.
Experience shipping on a sync engine such as Zero, Replicache, or ElectricSQL; the posting says this would be especially impressive.
Experience as a full-stack engineer at a data-heavy SaaS company with ownership of the warehouse.
Experience as a product engineer at a warehouse-native analytics or activation company, making Snowflake performant in an app.
Experience as a data engineer seeking to own the product feature that uses the data tables.

Responsibilities

Own the performance of first-party data in the app, enabling connected fields to be sorted, filtered, and charted without live warehouse queries.
Determine what data is precomputed, where it is stored, and how it stays fresh.
Deliver model scores—including propensity, LTV, churn, and product-fit scores—on person, segment, and product pages, with full distributions.
Own dynamic segments that update as new data arrives and have statistics ready for activation.
Own the in-app agent’s query tool using semantic views published by the data platform team, translating marketers’ questions into the appropriate data rows.
Own the product side of data synchronization, including precomputation decisions and the path from Postgres to the synced client.
Extend data pipelines by writing models or pipeline steps when a feature requires data structures that do not already exist.
Ship end-to-end features across Python AWS Lambda, Zero schema changes, and TanStack Start pages.
Build a repeatable system for delivering new scores and first-party tables to the app.
Make first-party fields sortable, filterable, and chartable without live warehouse queries, and provide the layer used by the in-app agent’s query tool.