Salesforce CI/CD pipeline with automated and AI code review by Adrián MariscalSalesforce CI/CD pipeline with automated and AI code review by Adrián Mariscal

Salesforce CI/CD pipeline with automated and AI code review

Portfolio project
Adrián Mariscal
I inherited a Salesforce deployment pipeline that let errors through and rebuilt it almost from scratch. Client name withheld under NDA.
How it works:
Every pull request from DEV to UAT runs three checks: PMD with custom rules based on Salesforce's recommended set, a validation of the package against UAT running all tests, and an AI model that compares the diff with the Jira task to flag out-of-scope changes or regression risk
If any check fails, the pull request is rejected and the developer gets the reason
If all pass, it is approved, merged and deployed to UAT automatically
From UAT to production the same checks run, with manual approval and deployment by the technical leads
Result: 25% less time from the end of development in DEV to the ticket being available in UAT.
Tech: Salesforce, Bitbucket Pipelines, PMD, Salesforce CLI, Jira, LLM-based review.

0 likes · 0 views · Nov 17, 2026

I rebuilt an inherited Salesforce CI/CD pipeline on Bitbucket: PMD, full test validation against UAT and an AI check of each diff against its Jira task. Time from DEV to UAT dropped by 25%.

Tags