About Nightshift
Nightshift is the data and analytics layer for AI agents. You connect a source once, write policy in plain rules, and any agent gets exactly the data you allow, enforced down to the row the moment it asks, served over MCP. We are a small, senior team in Pittsburgh and Miami. Our own agents run on Nightshift, governed by our own policies and our own audit log.
About the role
The connector is the front door. You will build the layer that registers a Postgres database, a Snowflake warehouse, or a Stripe-style API and turns it into a single governed catalog: schema introspection, type mapping, incremental sync, and a live view that never goes stale. No ETL, no second copy of the data, no place for a credential to leak. When this works, a customer points us at a source in the morning and their agents are reading governed data by lunch.
Representative projects
- Add a warehouse connector that introspects thousands of tables and presents one clean catalog, in minutes, without copying a byte
- Design the sync model that keeps the catalog honest against a source that changes under you
- Build the abstraction that makes the next connector a config file, not a fork
- Make a flaky third-party API feel like a first-class source: retries, backpressure, and an error a human can actually read
You may be a good fit if you
- Have built data integration, CDC, or connector platforms and know where the bodies are buried
- Are fluent in SQL engines and the reality that every database lies about something
- Care about correctness under partial failure more than about happy-path throughput
- Want to own a surface end to end on a team small enough that you can