Rosetta IMS
Software Engineer, Procurement & Inventory
Procurement, inventory and catalogue services for AlgoGroup that unify Shopify, retailer, supplier and spreadsheet data into one typed model, with validators that catch inconsistent costs and margin anomalies before they reach the live catalogue.
| When | Apr – Sep 2026 · via Talent Co-Op |
| Stack | Python, FastAPI, PostgreSQL, SQLAlchemy, Pydantic, Prefect, React |
| Links | github.com/kali-physi-hacker/rosetta-ims |
| 4,000+ | SKUs |
| ~60 | suppliers |
| 183 | automated tests |
A retail catalogue is only as good as the worst spreadsheet feeding it. AlgoGroup's procurement data lived across Shopify, retailer feeds, supplier price lists and spreadsheets, each with its own idea of a unit, a price and a margin. Rosetta IMS is the set of services I designed and built to bring that into one typed model.
One model, many sources
The catalogue covers more than four thousand SKUs from roughly sixty suppliers. Each source gets a typed ingestion path built on FastAPI, SQLAlchemy and Pydantic, orchestrated with Prefect, so that pricing, packaging units, bulk-buy terms, landed cost and margin all land in the same shape regardless of where they came from.
Validators at the door
The pipelines are backed by 183 automated tests, and the validators are the product. They flag inconsistent costs and margin anomalies before a change reaches the live catalogue, which turns a silent pricing mistake into a reviewable exception.
Provenance
Every catalogue change carries its provenance and an audit trail, so a number in the catalogue can be traced to the source, the transformation and the person or job that applied it. Reproducibility was a requirement, not a nice-to-have: procurement decisions get questioned, and the system has to be able to answer.
Next