Basis equips accountants with a team of AI agents to take on real workflows. We have hit product-market fit, have more demand than we can meet, and just raised $34m to scale at a speed that meets this moment.
Requirements
- We build the agentic ML systems that power Basis’s AI Accountant—so it can read documents, reason over context, and complete real accounting workflows safely and accurately.
- We’re practitioners of the new AI paradigm: rather than just tuning a model, we optimize the system around it—tools, memory, retrieval, orchestration, and evaluation.
- We push model providers to their limits when necessary (custom runtimes, bigger containers, nonstandard packages) and run the experiments required to learn quickly.
- We work from first principles with tight loops alongside Research, Product, Platform, and Accounting SMEs.
- We think in systems and care deeply about observability, clear abstractions, and code that’s easy to reason about in production.
- You’ll drive technical direction, architect systems, and review critical code
- Coach engineers not just to build better models, but to think better about systems.
Responsibilities
- Define and evolve our multi-agent architecture: autonomy boundaries, orchestration logic, context management, and safety layers.
- Own evaluation infrastructure—offline, online, and hybrid—that lets us ship models with confidence and traceability.
- Integrate retrieval, memory, and context management into production-grade agent loops; ensure stability under real workloads.
- Align closely with Research, Product, and Platform to translate insights into production systems with measurable impact.
- Insist on clean abstractions, legible systems, and deep observability; make complexity visible and manageable.
- Set and uphold high standards for experimentation, documentation, and decision quality.
- Continuously improve team processes—reviews, onboarding, retros, performance cycles—to compound speed and quality.
Other
- Build and lead the applied-ML organization
- Hire and grow a world-class team of ML and systems engineers; set crisp goals and coach continuous development.
- Foster a culture of rigor, iteration, and shared learning—where people move fast and stay grounded in reality.
- Establish clear processes for experimentation, evaluation, and delivery; make success criteria objective and comparable.
- Be a source of clarity and calm when things are ambiguous or hard.
