You will get your Salesforce accounts, contacts, or leads deduplicated and cleaned up
Project details
Duplicates are not a data problem, they are a trust problem. Once your team stops believing the account count, they stop believing the pipeline.
I have cleaned up roughly 1,100 confirmed duplicate account pairs with explicit winner and merge logic, without breaking the automation sitting downstream of those records. The hard part is never the merge. It is deciding which record wins, what happens to the child records, and which triggers you need to suppress while it runs.
Nothing merges until you have approved the rules and seen the counts, and every run ends with a reconciliation report showing exactly what changed.
I have cleaned up roughly 1,100 confirmed duplicate account pairs with explicit winner and merge logic, without breaking the automation sitting downstream of those records. The hard part is never the merge. It is deciding which record wins, what happens to the child records, and which triggers you need to suppress while it runs.
Nothing merges until you have approved the rules and seen the counts, and every run ends with a reconciliation report showing exactly what changed.
What's included
| Service Tiers |
Starter
$600
|
Standard
$1,800
|
Advanced
$3,200
|
|---|---|---|---|
| Delivery Time | 3 days | 7 days | 10 days |
Number of Revisions | 1 | 2 | Unlimited |
Frequently asked questions
About Gerry
Salesforce Developer | Apex, Flows, MCP & AI Integration
Reading, United States - 2:09 am local time
The technical core: Apex (triggers, batch and schedulable jobs, bulkified patterns), Flow architecture and the hard debugging (recursive loops, async paths, governor and scheduled-path limits), validation rules, permission sets, and integration users. When automation misfires in production, I find the actual cause instead of patching symptoms.
Where I’m different: I build AI-native tooling on top of Salesforce. I’ve built a custom MCP (Model Context Protocol) server in TypeScript with PKCE OAuth to let LLMs read and safely mutate Salesforce data, connected multiple orgs to AI assistants, and written CLI tooling to make org changes safer and faster. If you want to wire Claude or another LLM into your Salesforce workflows without handing it a loaded gun, that’s the work I do.
A few things I’ve fixed or built:
• Traced a production Flow that was hitting scheduled-path limits to a Task automation silently re-triggering itself, and stopped the loop without losing data.
• Killed trigger recursion in an account-hierarchy merge using a before/after-delete pattern with static-variable guards.
• Cleaned up ~1,100 confirmed duplicate account pairs with clear winner/merge logic, without breaking downstream automation.
• Built a Sailthru-to-Salesforce sync on schedulable Apex, abstracted through Custom Metadata so config changes don’t require code changes.
• Stood up an MCP server that lets an LLM query and update Salesforce through scoped OAuth, with explicit read vs. write boundaries.
What I build:
• Apex development: triggers, batch/schedulable, async, test coverage
• Flow design and debugging, including the ones that broke in production
• REST/SOAP integrations and middleware between Salesforce and your stack
• MCP servers and API tooling that connect LLMs to Salesforce safely
• Data migration, deduplication, and org cleanup
• CLI tools and scripts that automate the tedious admin work
How I work: I start by understanding the problem before I touch the org, because the fastest fix is often not the one you asked for. I write code that’s bulkified, tested, and documented, so your team isn’t stranded after the engagement ends. I scope honestly, flag risk early, and will tell you when a problem doesn’t need custom code at all.
If you’ve got an integration to build, a Flow that’s misbehaving, a migration you’re nervous about, or an idea about putting AI to work in your Salesforce org, send me the details and I’ll tell you honestly what it’ll take.
Steps for completing your project
After purchasing the project, send requirements so Gerry can start the project.
Delivery time starts when Gerry receives requirements from you.
Gerry works on your project following the steps below.
Revisions may occur after the delivery date.
Profile and plan
I profile your data and show you the duplicate population with match confidence.
Agree the winner rules
We agree together on which record wins and what happens to child records before anything merges.