A registered Claude Partner Network companySpreadsheet clean-up and merging
Messy customer or product lists deduplicated, standardised, and merged into one clean sheet.
Example project
- Rules for duplicates and formats, agreed with you
- One clean master list, originals untouched, with a report of what was merged
- A repeatable process for the next export
The problem
Three customer lists disagree: an Excel export from the old system, the sales team's Google Sheet, and a Mailchimp export. Duplicates, five phone number formats, names in different orders. Nobody trusts any of them enough to load into a CRM, so the CRM waits.
How the solution works
Illustrative example
Click each stepMerge the clear duplicates and keep the uncertain ones
Your team gets a cleaner list and a record of what changed, with the originals preserved.
Two exports describe the same customer differently
One sheet says A. Patel and another says Anna Patel. Both have the same customer ID, but another Anna Patel has a different ID.
- Inputs
- Your two customer exports
You agree which identifiers prove a duplicate and which source wins.
Apply the agreed matching rules
Claude groups the rows sharing the customer ID and standardises their formatting. The other Anna Patel is kept separate.
- Match evidence
- Same customer ID, not just a similar name
Rows the rules cannot settle are held for a person.
A clean sheet with a change report
Your team reviews the merged row, the separate customer and any unresolved records before using the new master sheet.
- Retained
- Original exports and an explanation of each proposed merge
Nothing is deleted from your original files.
Rules agreed first
What counts as a duplicate, which source wins, what a clean phone number looks like. Written down with you.
Claude merges
One master list, formats standardised, duplicates resolved by the rules. Anything the rules cannot settle is left for a person.
You review the report
What was merged, what was left, and why. Originals untouched, nothing deleted.
The process is repeatable
When the next export arrives, run it again.
What you get
- One list you trust enough to load into a CRM
- A report of every merge and every row left for you
- Originals untouched
- Run it again when the next export lands
Good fit
- Two or more lists that disagree and a CRM or mailing tool waiting
- A few thousand rows or more
Not the first project
- A few hundred rows; an afternoon in Excel does it
- The lists are already in one system
Engineers who take this on
0 of 3 selected. Choose up to three engineers for one brief.
December 1, 2026, 2 days a week
Software Engineer passionate about working on large distributed systems building web services.
Work example
Search
Engineered a high-performance Python and PostgreSQL search engine utilizing advanced keyword matching and optimized query understanding for transaction histories, help articles, account information and navigation links.
Claude Certified Architect | Hands-On AI Builder | Agents, MCP & Workflows That Ship to Production at Scale
Taking small projects now, part-time; more availability from December.
Senior backend engineer building Claude automations that connect securely to the tools you already use.
Work example
Client data imports that stopped breaking
At a retail-planning SaaS startup, client data arrived in many shapes and batch loads into the platform kept failing. I led the design of an extensible Python ETL framework that automated transforming and loading client data, cutting batch load errors by 75%.
Taking new work from November 2nd, up to 2 days a week (more flexible starting January 2027)
I help small businesses turn ideas and prototypes into secure, reliable software, at a speed made possible by Claude.
Work example
De-vibing a startup's codebase
I reworked an early-stage startup's AI-generated codebase for security and maintainability: removing hardcoded keys, adding row-level security to databases, building unit and end-to-end test coverage, and restructuring the code. I also consolidated the project's rules and Claude instructions to cut token usage. The founder now has a codebase they can safely keep building on.
Work example
AI that turns meeting notes into product specs
Small product teams were losing decisions and action items buried in meeting notes. As co-founder and sole engineer at Una AI, I built a pipeline that converts meeting transcripts into structured specs with decisions and action items, testing it against real meetings until the output was reliable. Design partners paid to use it.
Your brief is already filled in
Small · $2k to $5k
Free introductions by email. You agree scope and price with the engineer.