Founding AI Engineer

Andera; San Francisco (CA); On-site; USD 180,000 - 260,000; Full time

Job summary

Andera is building reliable AI agents for back-office finance, starting with automating audits for the world's largest public companies. You will help create core agent systems that enable long-horizon reasoning over huge data, ensuring reliability and scalable performance.

This high-ownership role spans architecture to deployment, collaborating closely with auditors and engineers to push fast, robust improvements across complex financial workflows.

Qualifications

• Experience building complex AI agent or workflow-based systems including planning, tool use, memory, and failure modes.
• Strong fullstack engineering ability with fault-tolerant end-to-end features.
• Systems engineering fundamentals for concurrent workloads on large datasets, with attention to throughput and observability.

Responsibilities

• Design and implement complex AI agent systems (planning, memory, tool use, recovery).
• Build fault-tolerant, end-to-end features integrating with the codebase.
• Process large data workloads efficiently across processes and threads.
• Debug production agent failures, edge cases, and performance bottlenecks.
• Rapidly learn new audit workflows and push agent improvements quickly.

Skills

AI systems

Fullstack engineering

Distributed systems

Performance optimization

Audit workflows

Tools

TypeScript

Python

Temporal

Docker

Terraform

AWS Fargate

Job description

About Andera

Andera is building reliable AI agents for back-office finance, starting with automating audits for the world's largest public companies.

Audits have resisted automation for decades due to the scale and instability of the underlying data. Large companies generate hundreds of millions of tokens of financial data across massive spreadsheets with constantly changing formats. Even understanding a single document can take expert auditors six or more hours, and a typical audit involves many of them.

We're building agents that can parse, interpret, and reason over data of this complexity—compressing gigabytes of messy financial information into representations that LLMs can work with reliably. The result is turning weeks of manual audit work into minutes of review.

This work requires long-horizon reasoning, robust data systems, and production-grade reliability—and it forms the foundation for automating some of the most complex financial workflows in the world.

Founding Team

Andera is built on the belief that this problem can only be solved by pairing elite auditors with elite engineers.

• Engineering team: Alumni of MIT, Berkeley, and Waterloo, with experience at Jane Street, Stripe, Shopify, and Microsoft.

• Audit team: Former #1 performers at Deloitte and leaders of national finance practices at Revolut.

The Role

We're hiring an AI Engineer to help build the core agent systems that power autonomous financial audits. You'll work on long-horizon reasoning, large-scale data ingestion, and production-grade agent reliability, collaborating closely with auditors and other engineers.

This is a high-ownership role with responsibility from architecture through deployment.

What You'll Work On

• Designing and implementing complex AI agent systems (planning, memory, tool use, recovery)

• Building fault-tolerant, end-to-end features that integrate cleanly with the broader codebase

• Splitting and executing workloads that process ~100GB+ of data across processes and threads efficiently

• Debugging agent failures, edge cases, and performance bottlenecks in production

• Rapidly learning new audit workflows and implementing agent improvements in hours, not weeks

What We're Looking For

• Experience building complex agent or workflow-based systems — You can reason about how to rebuild systems like Claude-style agents from scratch. You understand planning, tool use, memory, and failure modes.

• Strong fullstack engineering ability — You can be trusted to ship fault-tolerant features end to end. You understand how your changes interact with the rest of the system.

• Strong systems engineering fundamentals — You can design concurrent workloads over large datasets without stalling or failure. You think carefully about throughput, backpressure, retries, and observability.

• Comfort learning non-engineering domains quickly — You can absorb audit concepts from domain experts and turn them into working systems fast.

• Experience with or adjacent to our stack: TypeScript, Python, Temporal, Docker, Terraform, AWS Fargate