AI platform that gives accounting firms a full team of AI assistants
Member of Technical Staff - Atlas
You'll spend your mornings tracing how a multimodal agent parses a client's messy general ledger, and your afternoons deciding whether that logic belongs in a Python service layer or a TypeScript orchestration graph. This is backend-meets-full-stack work at the point where LLM pipelines meet actual money—reconciliations, tax workflows, the unglamorous core of accounting. The team ships agentic features that accountants trust with real client dollars, not demos.
What they're looking for
- 3-8 years shipping production systems in Python, TypeScript, or both—ideally bridging data-heavy backends and user-facing surfaces
- Experience designing around LLM failures: hallucination guardrails, retry logic, cost controls, human-in-the-loop fallbacks
- Comfort owning features end-to-end, from schema design through to the accountant's dashboard
- Track record in B2B or regulated domains where 'roughly right' isn't close enough—finance, accounting, or adjacent
- On-site in New York; this team hashes out agent architectures in person, not in threads