About this role
HubSync builds the platform the largest CPA and advisory firms in the country run on. We have an AI function that ships: agents running against our own systems, models wired into our internal operations platform, MCP services connecting them to real data, and an engineering org where AI code review and AI-driven QA are part of how software gets built. We move quickly and we would rather ship a working thing this week than design a better one for next quarter.
This role scales the internal half of that work. You will embed with a department at a time, starting with go-to-market, learn how the work actually gets done, and then build the agents and internal tools that take the manual parts away. This practice is growing, and you would lead it.
What you'll do
We judge the first 90 days on shipped things, not plans.
What this role is not
We are an equal opportunity employer. All applicants will be considered for employment without attention to race, color, religion, sex, sexual orientation, gender identity, national origin, veteran or disability status
HubSync builds the platform the largest CPA and advisory firms in the country run on. We have an AI function that ships: agents running against our own systems, models wired into our internal operations platform, MCP services connecting them to real data, and an engineering org where AI code review and AI-driven QA are part of how software gets built. We move quickly and we would rather ship a working thing this week than design a better one for next quarter.
This role scales the internal half of that work. You will embed with a department at a time, starting with go-to-market, learn how the work actually gets done, and then build the agents and internal tools that take the manual parts away. This practice is growing, and you would lead it.
What you'll do
- Embed with a department and diagnose, not intake. Go-to-market first, then finance, customer success, support, operations. Learn what people do all day, what they think the problem is, and what the problem actually is. The value you add over a request queue is telling someone "what you asked for saves two hours a week, but these three things instead change your quarter."
- Build and ship agents into production, fast. Gumloop, Relevance AI, n8n-class tooling, Claude and the other models we run on Bedrock, MCP into our own systems. The cadence is the point: sit with someone Tuesday, put something in their hands Wednesday, iterate with them Thursday.
- Build light internal tools when a tool is the right answer. A small web app, a dashboard, a form that writes to the right system. We care that it works and that people use it, and this is where being technical beats being a workflow configurator.
- Make each thing stick before you move on. Adoption, not delivery. An automation nobody runs is worth less than the conversation that produced it, and you will know the difference because you sat with them.
- Turn it into a practice. What you build, who owns it, what it saves. That becomes the playbook the rest of this team works from.
- Grow the team. You will shape the practice that two or three people run inside a year, and we will look to you for who they should be.
We judge the first 90 days on shipped things, not plans.
- By 30 days: multiple real friction points solved for real people, at least one agent running in production that someone uses without you present, and a working relationship with your first department.
- By 60 days: two or three departments in flight, a couple of internal tools live, and people coming to you rather than you chasing them.
- By 90 days: a repeatable motion for how you scope, build, hand off, and measure, and enough hours given back that finance can see them.
- You build with low-code and no-code AI tooling, and you build a lot. Gumloop, Relevance AI, n8n, Zapier-class orchestration, agent frameworks, MCP. What matters is not the logo list, it is that things you built are running and other people depend on them.
- You can code enough to be dangerous. Light web apps, scripts, APIs, and using AI coding tools to get there fast. You do not need to be a platform engineer, and you do not stop at the edge of a no-code tool.
- You are genuinely good with internal customers. You can sit with a finance manager or a sales leader, earn their trust in one meeting, ask the question that reframes their problem, and turn the answer into something technical. Consulting instinct, without the deck. This is the hardest part to teach and the part we screen hardest on.
- You move at an uncomfortable speed and you are calm about iteration. V1 tomorrow, better Friday, killed next month if it turns out to be the wrong thing.
- You think about the business, not just the request. Revenue, sales, finance, and customer operations: you have opinions about how these functions work and where they leak time.
- You are careful with the data. We work with CPA firms and their clients' tax information. Some of what you build touches systems that hold it, some of it must never leave our boundary, and knowing which is which is part of the job.
What this role is not
- Not IT, not enablement, not a training function. You build things. Enablement happens because what you built is good.
- Not a strategy or center-of-excellence seat. No frameworks, no maturity models, no governance council. Ship, sit with people, ship again.
- Not internal engineering tooling. AI code review, QA agents, and CI gates sit with a sibling team pointed at engineering. Your customers are every other department.
We are an equal opportunity employer. All applicants will be considered for employment without attention to race, color, religion, sex, sexual orientation, gender identity, national origin, veteran or disability status
Originally posted on