Odoo ERP & demand forecasting for logistics, manufacturing, and retail in the Czech–German–Polish triangle.
We are an Odoo Official Partner based in the Czech Republic. We replace fragmented ERP stacks with Odoo, and we bolt a production-grade forecasting engine onto the top of it. This GitHub organisation hosts the open-source components of that engine — the parts we ship into customer environments and maintain in the open.
→ datastruct.tech — company site, case studies, contact → Book a 30-min technical diagnostic — free, no slides → info@datastruct.tech
| 94% | Forecast accuracy on real customer SKUs |
| 84% | ERP cost reduction (SAP → Odoo migration, industrial customer) |
| 8 weeks | Average Odoo go-live — no 18-month projects |
Full case studies on datastruct.tech:
- SAP to Odoo in 14 Weeks — 84% Cost Reduction
- Czech Grocery Chain Cut Milk Spoilage by 40%
- On-Time Materials: From 67% to 94%
- Demand Forecasts for Shift Planning
These are the tools our commercial forecasting engine is built on. They are licensed permissively so you can adopt them without talking to us — but we'd be happy to help if you do.
| Repo | What it is |
|---|---|
| parqcast | Zero-computation data pipe from Odoo to Parquet. Extracts sales, inventory, purchasing, and warehouse tables as Parquet files ready for any planning engine. Odoo 18 + 19. |
| foreqcast | Linear-regression demand forecaster that reads parqcast exports and writes min/max replenishment rules back to Odoo stock.warehouse.orderpoint. |
| parqcast-lambda | AWS Lambda handler + S3/MinIO transport for running parqcast serverlessly. |
| parqcast-server | HTTP Parquet receiver with API-key auth and date/UUID-partitioned storage — the self-hosted counterpart to the Lambda path. |
Both Odoo addons ship with UI translations in 19 European languages (CZ, DE, PL, plus Nordic, Baltic, Mediterranean, and Arabic) — we operate across the CEE/DACH corridor and wanted the addons to feel native wherever they land.
Most forecasting vendors hand you a black box. Most Odoo partners hand you a consulting invoice. Our bet is that the boundary — how planning data actually leaves your ERP, and what gets written back — should be inspectable, debuggable, and replaceable. That's what these repos are.
- Founder: Oleg Popov — MIT Supply Chain Management '21, ex-Takeoff Technologies (automated micro-fulfilment).
- Recognition: Deloitte Technology Fast 50 (Central Europe).
- Where: Prague, Czech Republic. We operate in CZ, DE, and PL.
- Commercial enquiries: info@datastruct.tech
- Technical diagnostic (30 min, free): cal.com/oleg-popov-sjwko9/30min
- Issues & PRs on any of the repos above are welcome.