# At Vori, the forward deployed engineer is really an ops data engineer

> The listings disagree on whether the role needs three years' experience or five, but the job itself is mapping, reconciliation and data pipelines.

Original: https://fdetimes.net/en/analysis/vori-forward-deployed-data-engineer/

Brandon Hill, Vori's chief executive, sums up what independent grocers struggle with as a question: how do you close the gaps so people stop spending 60 to 70 per cent of their time moving data "from left to right"? The company's answer, judging by its Forward Deployed Engineer listing, is to hire someone to close them.

Vori has 29 employees, went through Y Combinator in the W20 batch and calls itself "the modern operating system for grocery". The FDE role is posted in two places. On YC it is called Forward Deployed Data Engineer: full time, in San Francisco, with visa sponsorship. On Built In it is called Founding Forward Deployed Engineer, with a more detailed job description.

Read closely, this is not a job building AI demos in a client's office. It is data engineering, done from inside the operations team. For anyone aiming at FDE work, listings like this one show fairly clearly which skills companies are actually paying for.

## An operating system is only useful once the customer's data is in it

Vori presents itself as an all-in-one operating system for the food industry, with a Back Office product aimed at shrink (lost or unaccounted-for stock) and operational efficiency. But an operating system is useful only if it holds the store's own data: products, suppliers, cost prices, quantities received. Without them, the product is just a nice interface.

That makes configuring VoriOS for each customer a data project at heart. The Built In listing asks candidates to understand VoriOS end to end, including how the system and its integrations depend on each other.

Knowing one module well is not enough. A bad mapping at goods-in becomes a wrong number in the shrink report, and the FDE has to be able to trace that path.

**Key point:** When you sell an operating system to grocery stores, deployment means getting the store's data into it.

## One invoice through the FDE's hands

Picture an independent grocer that has just signed. Each week the store receives dozens of distributor invoices, mostly as PDFs or photos. One invoice line might read "CHBN GRK VAN 5.3Z 12CT", while the point-of-sale catalogue says "Chobani Greek Vanilla 5.3 oz".

The first job is extraction, the area the listing names outright: unstructured data, ETL pipelines and ML/OCR. Next comes normalisation.

Regex splits the pack size "12CT" into 12 units per case and "5.3Z" into 5.3 ounces. SQL then joins the cleaned line to the store's product catalogue.

SQL, regex and scripting are also the toolkit the listing names.

Reconciliation comes last. Say the invoice shows 48 cases but the receiving slip shows only 46. The 2-case gap, or 24 pots of yoghurt, has to be flagged before it disappears into the month-end shrink figure.

Hill's figure shows how big this work is. Take a store manager working 50 hours a week: 60 to 70 per cent of 50 hours is 30 to 35 hours spent copying data from one system to another. Each invoice type an FDE automates hands some of those hours back to the store.

## Inside the operations team, not a separate services unit

The most telling detail is where Vori puts the role in the organisation. The listing describes configuring and maintaining mappings and reconciliation as an integral part of the operations team. On that reading, the FDE does not finish a project and hand it over, but keeps the mappings right every time the customer's data changes.

The FDE also has to work closely with Operations, Product and Engineering to improve processes and systems end to end. The design has an obvious strength: the person who first sees a mapping error is also in a position to propose a product fix so the error does not recur at the next customer.

The risk is the other side of that. Sitting in operations can easily turn into fixing data by hand indefinitely. Anyone who takes the role should start measuring the share of lines mapped automatically from the first week and track it as it rises. Otherwise the "left to right" problem has simply moved from the store to Vori.

## Three years or five? Where the listings contradict each other

The top of the YC listing says 3+ years of experience. The Built In listing asks for 5+ years in FD or Solutions Engineering, or a similar customer-facing technical role. For a 29-person company hiring a "founding" role, the gap suggests years of experience are not a hard cut-off.

What the Built In listing stresses more is autonomy: candidates must handle extraction, ingestion and pipeline tuning with very little guidance. Alongside that comes the ability to work directly with customers' IT teams, who know how to export data from the store's old systems.

The table below matches each line of the listing to the actual work in a store and to how to show that skill to a hiring manager.

| Line in the listing | Actual work in the store | How to show it on a CV |
|---|---|---|
| Unstructured data, ML/OCR pipelines | Reading scanned invoices and photographed delivery slips | An OCR project on real documents, with its extraction accuracy |
| SQL, regex, scripting | Splitting pack sizes, joining to the product catalogue | Concrete examples of normalisation rules you wrote and how many lines they handled |
| Mapping and reconciliation within operations | Flagging gaps between invoices and goods received | Number of data sources mapped, automation rate before and after |
| Working directly with customer IT | Getting a correctly formatted export from a legacy system | A time you sorted out an integration with the customer yourself, without a go-between |
| Understanding VoriOS end to end | Tracing a wrong report figure back to the original invoice line | An incident you traced back across several systems |

## Describe a messy mapping job, not your years

At a company selling an operating system to grocers, ingestion quality decides whether the product works at all. If you have 3 or 4 years of experience and have owned a pipeline yourself, do not skip listings that ask for 5.

A bullet such as "mapped 6 supplier data sources, raising the automatic match rate from 40% to 85%" is worth more than a line summing up your years. The numbers in that sentence are only a template; put in your own.

When interviewing at companies like this, be ready to walk through one messy mapping job in detail: what the raw data looked like, which rules you wrote, which errors slipped through, and whom you spoke to on the customer's side to fix them.

Hill's question already points to how the role will be judged. At Vori, a good FDE is measured by the hours a store manager no longer spends copying data.

**Try this week:**

- Take a few public PDF invoices or delivery slips and build a small pipeline that runs OCR, splits lines with regex, then joins them to a sample product catalogue with SQL
- Measure the share of lines that map automatically and log each type of error that remains. That is the number to put on your CV
- Rewrite one CV bullet to this pattern: which customer data source, how much you automated, and whom you worked with on the customer's side

## Sources

- [Forward Deployed Data Engineer at Vori (Y Combinator)](https://www.ycombinator.com/companies/vori/jobs/trpOGJ5-forward-deployed-data-engineer)

- [Founding Forward Deployed Engineer (Built In)](https://builtin.com/job/senior-solutions-engineer/6923600)

- [Exclusive: AI grocery startup Vori raises $22 million to help independent retailers compete with Walmart and Amazon (Fortune)](https://fortune.com/2026/05/05/exclusive-ai-grocery-startup-vori-raises-22-million-to-help-independent-retailers-compete-with-walmart-and-amazon/)

- [Funding News (Vori blog)](https://www.vori.com/blog/series-a)
