Founding AI Engineer

MIRIAN AI is building the agentic financial operating system for PE-backed mid-market companies. Our AI agents don't just track tasks — they do the actual accounting work: computing accruals, reconciling transactions, generating journal entries, and improving with every close cycle. Humans review and approve.

We're looking for a Founding AI Engineer who can design and build production agentic systems from scratch — not wrappers around ChatGPT, not chatbots, not prompt playgrounds. Real multi-step AI pipelines that handle financial data with precision.

What you'll own:
Agentic workflow architecture — multi-step LLM pipelines with tool use, memory, self-correction, and human-in-the-loop checkpoints
LLM orchestration — prompt engineering, structured output parsing, retrieval-augmented generation, evaluation frameworks
Financial reconciliation and accrual computation agents that learn from historical patterns
Integration with Anthropic Claude APIs, Langfuse observability, and Temporal workflows
Defining how AI gets applied to real accounting workflows — not theoretical, production

Stack: Python, FastAPI, PostgreSQL, Temporal, Anthropic Claude APIs, Langfuse, AWS

You should have:
4-5+ years software engineering experience, with at least 1-2 years building agentic or multi-step LLM systems in production
Deep understanding of tool-use patterns, chain-of-thought reasoning, retry/self-correction loops, and structured output validation
Strong Python backend skills — you're not just an ML researcher, you ship production code
Experience with evaluation and observability for LLM systems
Ownership mindset — you figure out what needs to happen and do it

Bonus points:
Fintech, accounting, or ERP domain knowledge
Prior 0→1 startup experience

Why MIRIAN:
Founding-team ESOPs and impact — shape the product and architecture from day one
$5B+ market dominated by legacy tools with zero agentic AI in production
Remote-friendly, flexible hours