Member of Technical Staff

Sixtyfour, San Francisco, California

What you will do

OSINT agents at the core of the product: how they plan an investigation, which sources they check, how they settle conflicting evidence, and when they decide they're done.
New data sources and tools for the enrichment engine.
Evals that tell us whether a prompt, model, or agent change made results better or just different.
Batching, caching, retries, and orchestration for jobs that enrich millions of records at a time.
Features end to end, from design doc to production.

Who we're looking for

Strong fundamentals. You understand concurrency, networking, databases, and APIs below the level of whatever library you're calling.
Design judgment. Given a vague problem, you can break it apart, find where it will fail, and defend the design you pick.
A habit of thinking about scale. You consider data volume, latency, concurrency, and cost before anyone asks, and you can point to systems where you had to.
Something real you've shipped (a product, an open-source project, research, a hackathon build) that you can walk through decision by decision.

Sixtyfour, San Francisco, California

OSINT agents at the core of the product: how they plan an investigation, which sources they check, how they settle conflicting evidence, and when they decide they're done.
New data sources and tools for the enrichment engine.
Evals that tell us whether a prompt, model, or agent change made results better or just different.
Batching, caching, retries, and orchestration for jobs that enrich millions of records at a time.
Features end to end, from design doc to production.

Who we're looking for

Strong fundamentals. You understand concurrency, networking, databases, and APIs below the level of whatever library you're calling.
Design judgment. Given a vague problem, you can break it apart, find where it will fail, and defend the design you pick.
A habit of thinking about scale. You consider data volume, latency, concurrency, and cost before anyone asks, and you can point to systems where you had to.
Something real you've shipped (a product, an open-source project, research, a hackathon build) that you can walk through decision by decision.