Data Engineer (Financial Automation & Dashboards)

TL;DR
The CFO Group is a fractional CFO firm serving founders with $5M to $100M in revenue. What we actually sell is visibility: a rolling 13-week cash forecast, a live dashboard the founder checks instead of ignores, and a real financial model behind every major decision.

Producing that today depends on people moving data by hand. Every week someone exports files from a client's accounting system, CRM, payroll platform, and bank, normalizes formats that never agree, reconciles the numbers, and rebuilds the reporting. We are hiring a data engineer to replace that with pipelines, and then turn the whole thing into a stack we can deploy to every client we take on.

You would be our first technical hire. The data infrastructure you build is the product.

Compensation: $5,000/month base plus up to $1,500/month performance bonus ($6,500/month on target). Full-time, fully remote. Eight hours a day, with only a two-hour required overlap: 8:00am to 10:00am US Eastern. You schedule the rest.

Why this role exists
Our clients do not have a numbers problem. They have a visibility problem. They are growing, profitable on paper, and still chasing payroll, because revenue is not cash and no one has built them a system that shows the difference before it hurts. Our whole value proposition is closing that gap: showing a founder where their cash is actually going, what their next hire truly costs, and which parts of the business make money.

The obstacle is that almost everything required to produce that visibility is manual data work. A finance person logs into five systems, pulls five extracts that disagree with each other, cleans them in Excel, reconciles them against the bank, and rebuilds a report. It is slow, it is fragile, and it scales only by hiring more people to do more of it. It also means the numbers often land days after the founder needed them, which is precisely the problem we are supposed to be solving.

That work is now code's job. Extraction, normalization, reconciliation, validation, exception flagging, and increasingly the first draft of the variance narrative itself. Automate that layer and our CFOs spend their hours on judgment instead of data janitorial work. Our domain is cfogroup.ai for a reason, and we would like the product to catch up to the domain.

We have proven the model once. For an insurance services client whose data was spread across a CRM, an accounting system, a contractor payroll platform, and monthly commission statements from several carriers, we built a live executive dashboard covering revenue, unit economics, cost of sales, gross profit, and cash. Their CEO and board use it. It works.

But it is still hand-fed every week, and rebuilding it from scratch for every new client is not a business. Your job is to fix both halves: automate the pipeline, then make it repeatable.

What you will do
Build the ingestion layer. Scheduled pipelines that pull data out of client accounting systems, CRMs, payroll platforms, banks, and vendor portals. Some sources have clean APIs. Many have nothing, and will require parsing email ed PDFs and Excel files, or scripted browser automation against a portal that was clearly built in 2009. You will d ---------- the right approach per source and build it.

Model the data. Every source names things differently and none of them agree. You will design a normalized data model that becomes the single source of truth for a client, with all the vendor-specific ugliness absorbed into code rather than living in a person's head.

Make reconciliation automatic. The numbers have to tie. Transaction data to statements, statements to bank deposits, payroll to the P&L. You will build validation and reconciliation into the pipeline so breaks surface immediately, instead of surfacing in front of a founder or a board.

Serve the reporting layer. The 13-week cash forecast, the hiring model, the margin-by-segment view, the executive dashboard. These are what clients pay us for. You will make them live, accurate, and self-updating rather than a spreadsheet someone refreshes on a Sunday night. This means owning the presentation layer too, so you need to be comfortable in HTML and JavaScript, not only in the pipeline behind it.

Productize the stack. This is the part that matters most. The goal is not one bespoke dashboard. It is a platform we can point at a new client and stand up in days, using connectors we already built and a reporting layer that gets configured rather than rewritten. You make the second, fifth, and twentieth deployment dramatically cheaper than the first.

Use AI as a real tool, not a demo. We use AI coding assistants and agentic tools every day and expect you to. A genuine part of this job is finding where an LLM does in seconds what a person does in an hour: classifying transactions, extracting structure from unstructured documents, explaining variances, summarizing exceptions. We want someone already fluent with these tools and opinionated about where they belong in a system that must produce correct numbers, and where they absolutely do not.

Talk to clients. Not constantly, but you will join calls to scope out a client's systems, learn what their data actually means, and demo what you have built. You need to hold a conversation with a non-technical founder or controller without condescending or disappearing into jargon.

What we are looking for

Must have:
3+ years building production data pipelines in Python. Real ETL: extraction, transformation, scheduling, orchestration, error handling, the parts that are not glamorous.

Strong SQL and sound instincts for data modeling. You can look at four inconsistent source systems and design the schema that reconciles them.

Experience with genuinely hostile data sources: undocumented APIs, CSV and Excel exports, PDF parsing, web scraping, and browser automation when there is no other way in. This is most of the job, so tell us about the worst one you have beaten.

Enough front-end capability (HTML, JavaScript, a modern framework) to build and maintain the dashboard, not only the pipeline feeding it. You do not need to be a designer. You do need to ship a working interface.

Demonstrated fluency with AI coding tools and LLM APIs. We will ask what you have actually built with them.

Obsessive numerical accuracy. In finance, a pipeline that is 99% correct is a pipeline that is wrong, and the person who finds out is a founder in a board meeting.

Clear written and spoken English, and the judgment to be trusted on a client call.

Nice to have:
Exposure to accounting or finance: general ledgers, P&Ls, reconciliations, cash forecasting. You do not need to be an accountant, but a balance sheet should not scare you.
Cloud deployment, job orchestration and scheduling, disciplined version control.
Experience building internal tools for non-technical users who will never read your documentation.
A track record of automating something nobody asked you to automate.

Please note: this is not a QA or test automation role, and it is not an RPA role. "Automation" here means automated data pipelines and reporting, not Selenium or UiPath.

Who this is not for: someone who needs a fully specified ticket before they can start. Problems arrive here as "these two numbers do not match and nobody knows why." You have to run that down yourself, and you have to care that it is wrong.

Who you would be working with
The CFO Group was founded by Joe Isaac, who audited at Ernst & Young, spent a decade at Amazon including leading FP&A for its $250B International Retail segment, and served as a strategic finance partner on the Amazon and Apple retail collaboration. Our clients are founders in construction, contracting, retail, restaurants, health and fitness, and insurance. You would report directly to Joe.

This is the first technical hire at a firm betting its growth on automation. There is no legacy stack to inherit and no committee to route decisions through. If you build the right thing, it ships.

Compensation and logistics
Base salary: $5,000 USD per month.
Performance bonus: up to $1,500 USD per month, for $6,500 per month on target.

How the bonus actually works: it is tied to delivery milestones you control, not to company revenue. We set the targets together at the start of each quarter and write them down. Things like getting a client's data sources fully automated, cutting deployment time for the next client, or eliminating a recurring manual process entirely. If you hit what we agreed, you get paid.

Type: Full-time, fully remote.
Hours: Eight hours a day. The only fixed requirement is that you are online and available from 8:00am to 10:00am US Eastern, which is 8:00pm to 10:00pm in Manila (9:00pm to 11:00pm during the US winter, since the Philippines does not observe daylight saving). The other six hours are yours to schedule, along with your breaks. This is not a graveyard shift and we are not asking you to work one.

Benefits: Medical

Data handling: you will work with confidential client financial data and are expected to treat it accordingly.

How to apply
Send a short note covering:
1) A manual process you automated. What was it, what did you build, and what happened to the hours it used to consume?
2) The worst data source you have ever had to integrate, and how you finally got the data out of it.
3) One thing you have built using an AI coding assistant or an LLM API.

Please skip the generic cover letter. Three specific paragraphs beat a page of adjectives, and we will read them.