The opportunity

An engineering team in New York works as the data transformation partner to some of the largest private-equity-backed portfolio companies in the market. The mandate is the ugliest data work those businesses have: untangling systems after a merger, converting documents into clean structured records, retiring spreadsheet workflows in favour of real software, and holding customer and financial data in sync across tools that were never designed to talk to each other.

The delivery model is the unusual part. Engineers embed inside the customer's business, put a working prototype in front of them within 48 hours, and have production software running on real data in roughly four weeks.

The customer base is already there: a large majority of the biggest funds in the industry, plus the portfolio companies they own, which opens a direct line into hundreds of businesses.

What you would build

Custom applications, data pipelines, integrations, and AI agents, aimed at the problems that recur at nearly every portfolio company:

• Migration after a merger. Pull scattered contract, customer, and financial data into a single source of truth once two businesses become one.
• Document and data pipelines. Convert documents and systems of record into structured, cited output that downstream teams can actually rely on.
• Operational applications. Replace the paper, the binders, and the spreadsheets with software the team genuinely runs on.
• Revenue and CRM operations. Tidy the pipeline data, automate renewals and pricing, and keep customer records consistent everywhere they live.
• Integration and agents. Feed clean data into the ERP, the CRM, and the warehouse, and stand up agents that absorb the repetitive work.

What the job actually feels like

• You embed. Sit with the people doing the work, learn how the business really runs, and identify the problem that is worth solving.
• You prototype fast. Something that works within two days — not a deck describing something that might.
• You ship. Prototype to live software in about a month, running against their real data.
• You stay accountable. You remain on the hook until the thing works and is creating value, then hand over software the company owns outright.
• You carry it forward. What you build at one company travels to the next, so every engagement starts further ahead than the last.
• You are the face on site. Direct contact with founders, operators, and the deal and value-creation teams behind them.

Who you are

• A builder with mileage. Four or more years shipping production software, with genuine ownership of the results.
• Quick and technically strong. Python, SQL, APIs, and data pipelines, plus real comfort building with LLMs, retrieval, evaluations, prompting, and agents.
• Full-stack enough to finish. You can put a usable application in front of a non-technical team, not just stand up a backend.
• Sound in your judgment. You can take an ambiguous business problem, talk it through with operators and executives, and decide what deserves to be built.
• High agency. You show up on site, navigate a messy organisation, and make things work without waiting for a specification.
• New York based, and content to travel out to customer sites.

Visas are sponsored, any and all that the work requires.

How the process runs

• Three or four conversations with the founders and the team.
• A paid work trial. You take on a real problem and are paid for your time. It is mutual diligence on how you diagnose, scope, and build when the instructions are thin.
• A dinner, lunch, or coffee with the team before any decision is made.
• For the right person, first call to decision can happen in about a week.