Technical Cofounder — building AI native public accounting service (10-person big 4)

This is a cofounder role; not a paid advisor or CTO or engineer role. There's no cash compensation yet.

Please only apply if you want to build and are ready to build with another cofounder.

About the company We're building a ten-person Big 4 a SaS (service as softare) company: a licensed accounting and tax advisory firm where an always-on team of AI agents does the work and a small number of licensed CPAs review and sign.

Clients buy an outcome — a technical accounting memo, an R&D credit study, an income tax provision, a valuation — at a fixed price, delivered in 48 hours instead of three weeks. Public accounting is a $650B industry priced on the assumption that expertise is human and therefore scarce. That assumption just stopped being true.

Not software sold to accountants. A firm. Law has its AI-native firm. Medicine has one. Accounting doesn't yet.

What you'll own Engineering and product, end to end, as a genuine cofounder — not a first engineer

The agent stack: multi-agent systems that draft professional deliverables to a standard a licensed CPA will sign

Document ingestion and extraction from messy client source material — trial balances, general ledgers, prior-year returns, contracts, board minutes

A hybrid reasoning architecture: deterministic calculation where numbers must tie out, model reasoning where judgment is required

Source grounding and citation verification against primary tax and accounting authority

The review layer — self-checking, uncertainty flagging, and escalation, so human review drops from 100% of output to a fraction

Audit trail and evidence generation that holds up to a regulator, an insurer, or a peer reviewer

The learning loop: playbooks, client context and patterns that make engagement #20 dramatically cheaper than engagement #1 Technical hiring and architecture decisions as we grow

Requirements Founder intent 5+ years building software, with meaningful time as a senior/staff/lead engineer or a technical founder

Shipped production LLM or agent systems that real users depended on — not demos, prototypes, or internal experiments

Strong Python (or equivalent) and hands-on experience with modern agent frameworks, tool use, RAG and evals

Experience in correctness-critical domains where "close enough" fails — finance, healthcare, legal, compliance, payments, or similar

Comfortable with document-heavy and calculation-heavy problems (PDF/Excel parsing, structured extraction, numerical accuracy) Able to go from zero to a working, client-usable system independently and fast Comfortable operating pre-funding, pre-product, with ambiguity that resolves through customer contact rather than planning

Nice to have Built evaluation harnesses or QA systems for LLM output Experience with SOC 2, data security, or handling regulated client data Prior startup founding or very early-stage experience Familiarity with accounting, tax, or fintech workflows — a bonus, not a requirement

Who you'd be working withLotus Li, CPA — I've sat on every side of this industry: KPMG — senior associate, audit Razer (VP of Finance) and Printronix (Corporate Controller) — the buyer, hiring Big 4 and regional firms and knowing exactly which fees were worth paying

Praxis Business Solutions — five years running my own tax and advisory practice: R&D credits, tax provisions, transfer pricing, technical accounting, SEC reporting, 15+ clients, ~$3M revenue

Hive — three years building AI agents for tax research, tax-return analysis and multi-year planning, used by thousands of CPAs and financial advisors

Hive is also where I learned the lesson this company is built on: firms don't transform themselves. The transformation has to show up as a competitor.

Process Intro call — 45 minutes, both directions Deep dive — I walk you through the domain and the first engagement; you walk me through what you've built

Build together — one real client engagement, 1-2 weeks before either of us signs anything Paper it — if it's good, we do equity properly with a lawyer

Set alert for similar jobs