Want to know if this job is worth applying to?
Remote
Remote only
Exactera has offices in New York City, Tarrytown NY, San Diego, CA, London, and Argentina.
Exactera helps large multinationals handle complex tax work: transfer pricing, R&D tax credits, indirect tax, and audit defense. We are moving from a software-and-services model to an AI-native platform that does the work itself, with tax practitioners applying judgment where defensibility requires it.
This role owns the product direction for the agentic platform behind that shift: the system that enables AI agents and practitioner tools to run tax analysis on trusted, current data. The data foundation (a lakehouse, external sources, and the contracts that keep them reliable) is the substrate. The product is what agents and practitioners can do with it.
You work directly with the Principal AI Engineer, who owns the AI/ML architecture, and the Principal Data Engineer, who owns the data tier. You own what the platform does, in what order, and why. Because you define how the company’s practitioners and agents use AI, you have to work deeply in these tools yourself.
Exactera is growth-stage, so the role comes with equity and the room to set the platform’s direction yourself.
The six outcomes that define success in the first 18 months.
O1. Own the product strategy and roadmap for the platform.
You own what the platform is for and what it builds next. The roadmap ties to practitioner outcomes rather than feature counts. When priorities compete across engineering, product, and the practice teams, you decide and keep everyone working from the same plan.
O2. Define what compliance-grade means for this platform.
Tax work has to be defensible. You set the product requirements that make the platform’s output trustworthy: where results must be deterministic, what guardrails constrain agent behavior, how tool access is scoped, and what has to be auditable. You decide where human judgment stays in the loop. This is the bar that regulated tax work requires and general AI tooling does not clear.
O3. Own the agentic access surface.
The platform’s main product is the surface AI agents and practitioner tools use to reach data and act on it. You own that surface: the interface, the contracts behind it, and the limits on what agents can do. How this works decides how the whole platform gets used.
O4. Treat the data foundation as a product.
The data foundation is what the rest depends on, so you manage it as a product. With the Principal Data Engineer, you set the contracts between data producers and consumers: schema, quality, and freshness as commitments practitioners can rely on.
O5. Make external data a platform capability.
The platform is only useful if it reflects what happens outside it: regulatory changes, SEC filings, financial data, third-party sources. You decide which sources matter and how they show up in the product, weighed against their cost. Each is a coverage and value decision.
O6. Measure success by practitioner outcomes.
You measure success by what practitioners can do with the platform: faster, more accurate work with less of it done by hand. A pipeline in production that does not change a practitioner’s day has not succeeded.
Required
Preferred
