LMI Automation Pipeline
From 57% Data Loss to 5% — Built from Zero
The Problem
95% of LMI referrals for RWP and 15% for DNR/IM were not getting queued due to a manual macro process break. This meant thousands of weekly order referrals were being lost, representing an INR 816M annualized opportunity that was going unaddressed.
My Approach
I self-taught Python and AWS in three weeks to build the solution. The pipeline uses the Maxis API for data extraction, CRON jobs for JSON payload extraction and cleaning, and a Datanet script for delivery scan overlay. It automates ticket portrayal to ATS and 3P Trans teams, eliminating the manual macro dependency entirely.
The Outcome
Data loss reduced from 57% to 5%. An additional 13K orders per week are now referred for LMI. INR 374M annualized savings potential unlocked. 182 hours of annualized productivity savings for the operations team.
Why It Mattered
This project fundamentally changed how I approach analytics. I went from being someone who analysed data to someone who could build the infrastructure to capture it. It proved that an analyst with the right motivation can solve engineering problems — and that sometimes the fastest path to impact is building it yourself.
