Global spend-analytics platform
Own the end-to-end data pipeline for the company's largest client — spend data across 5 global regions and 13 sub-regions (~6M invoice rows plus PO, lookup and open-PO datasets; ~500–600k rows refreshed weekly).
Automating ingestion
Replaced a manual KNIME upload process with fully automated email-based ingestion (Microsoft Graph API → remote SQL Server batch script), and standardised file-submission and naming conventions so every one of the 13 regions feeds one consistent intake.
Data quality at scale
Built automated validation — row and hour counts, data-type and column-consistency checks — that drops stray columns, logs discrepancies, and emails client and internal teams for review before anything reaches a report.
Downstream
Vendor/spend lookups, multi-language data cleaning, an ~11–12k-rule categorisation engine, and advanced spend analysis (rolling / 4-week windows, PO splitting) feeding dashboards. Now consolidating with the client's IT and cybersecurity teams into a single unified pipeline.
Result
Cut weekly processing from ~4–5 hours of manual upload-and-check to a ~75-minute automated run — around 3–3.5 hours saved weekly — while removing manual number-checking for the highest-billing account.
Stack
Interested in work like this?
dhruvnirmal2111@gmail.comOpen to remote data & AI roles — Australia, India, and the Gulf / Singapore.