Founding Forward Deployed Engineer

Cranston is building the AI back office for CPA firms. This role turns messy tax prep, bookkeeping, client-request, and firm-operations workflows into shipped product inside live accounting firms.

You will get on calls with partners, preparers, reviewers, controllers, and firm operators; map how the work actually happens today; identify the edge cases; and use AI agents to turn that context into working software. The job is not to write specs from far away. The job is to understand the customer, build the workflow with agents, use it yourself, and keep tightening it until a real firm can trust it.

About the role

You will use AI agents as your primary implementation layer, but you still need product judgment and enough technical fluency to know when the system is right. You should understand how modern software fits together, be comfortable around GitHub and Postgres, and be able to reason about product behavior from logs, database state, screenshots, and customer feedback.

This is a customer-facing product engineering role. You will spend time with users, shape ambiguous operational pain into concrete product direction, drive Codex and other agents through implementation, and verify the finished workflow in the actual product. The best person for this role is high-agency, product-minded, technically fluent, and unusually good at turning ambiguity into a next move.

About Cranston

Cranston was founded by Sean O’Bannon and Max Minsker to rebuild the accounting firm back office from the ground up. Sean previously built ERP and automation systems at ReMatter and Databricks, seeing how much labor still sits on top of even the best systems of record. Max ran an accounting firm that filed 10,000+ tax returns and saw the same problem from inside the work itself: firms losing their best hours to document chasing, data entry, reconciliations, cleanup, and handoffs.

Our view is that the next generation of accounting firms will not run on bigger offshore teams or more disconnected point solutions. They will run on AI agents that can understand firm context, do the repetitive prep work, and leave the judgment calls to experienced accounting professionals. We are building that back office: the system that helps CPA firms take on more work, serve clients faster, and give accountants their time back.

What you will do

• Run workflow-discovery calls with CPA firms and finance operators, getting from vague pain to the exact steps, exceptions, documents, systems, and handoffs behind the work.
• Translate messy back-office operations into clear product direction for tax prep, tax review, bookkeeping, client requests, and practice-management workflows.
• Use Cranston's internal agent platform to package customer context into builds quickly instead of starting every feature from scratch.
• Steer agents through the implementation loop, review behavior in the product, and keep tightening the workflow until it is good enough for live customer use.
• Use the product yourself, inspect data and logs when needed, notice what is wrong, and push the system until the details are right.
• Help onboard customers onto the workflows you shipped and keep improving them from real usage.

What you will do in your first 30 days

• Sit in on customer and operator calls to learn how CPA firms actually move work from source documents to reviewed tax packages, books, and client follow-ups.
• Shadow existing Cranston workflows and identify workflows where better product behavior would remove real manual work.
• Turn that workflow into a concrete build plan: inputs, user-facing states, edge cases, success criteria, and verification steps.
• Drive an AI-agent implementation loop from customer context to working product.
• Test the workflow yourself against real or realistic data, fix the rough edges, and help ship it into a live customer or internal operator workflow.

Who we are looking for

• You are remarkable in some way that is hard to fake. You may have been unusually good at games, music, writing, sales, math, building internet things, or a niche obsession most people do not understand.
• You are product-minded, good at understanding systems, and unusually fast at turning ambiguity into a concrete next move.
• You do not need a CS degree, but you should be technically fluent enough to work effectively with AI coding agents and judge whether the output actually solves the problem.
• We care much more about raw ability, taste, curiosity, and output than traditional credentials.
• You are comfortable using AI coding agents as your main implementation tool and holding their output to a high bar.
• You are probably a Lovable, Replit, Cursor, Claude Code, or Codex power user. You have built apps, internal tools, automations, websites, or strange little products because you could not help yourself.
• You have a bias toward the simplest thing that could possibly work, especially when everyone else is making the problem sound more complicated than it is.

Interview Process

Our process is lightweight and practical. We want evidence that you can understand a messy workflow, turn it into a concrete product direction, and use agents to ship useful software.

Typical process:

1. Intro conversation about your background and what you have built.
2. Practical product/workflow exercise using a real or realistic back-office workflow.
3. Founder conversation focused on taste, pace, customer empathy, and how you work with AI agents.

We try to move quickly and keep the process tied to the actual work.