Build Lean MVP: Forecast & Inventory Planner (Python + React)
Budget: $4,000 – $8,000 USD
TITLE
Build Lean MVP: Forecast & Inventory Planner (Python + React) — Fixed Price US $7–8k
PROJECT DESCRIPTION
Context
We need a lightweight web tool to replace an Excel‑based forecasting spreadsheet system. The output feeds an existing distribution engine (out of scope here).
Scope – Phase 1 MVP
1. Data Upload & Mapping
- Drag‑and‑drop or SFTP file upload.
- Web UI (React + Tailwind) lets users map columns to target fields; template saved per customer.
- Ingest data to a Postgres or SQL Server database
2. Forecast Service
- Prophet/XGBoost (or similar) produces baseline demand and confidence bands using statistical methods for trend, seasonality, consistency, etc. using the supplied data and inferences from it
- Get data feed from mapped files, establish inventory levels and current movements (inbound, out, committed, etc) for each item
3. Nightly Files Export
- After each nightly run, generate “Binlevels_.csv” containing stock, inbound, outbound, SPL‑Min and shortage per SKU, as well as key forecasting data per item-location combination.
- Generate a suggested purchase orders file (based on item lead times and repair re-work lead times if applicable)
- Generate an overall inventory status file (with item and location attributes, stock levels, usage for past 6 months, and various other available info you'll concatenate)
- Generate a SLA-at-risk nightly file (where we believe we have no stock or are vastly understocked, to be expedited, and are very likely to have an incident based on forecast). Send this as an email with a comment.
- Drop the files to a designated network share or SFTP folder.
4. Portal + Chat What‑If
- User to login with a SSO
- Teams/Web chat (OpenAI o3 key supplied, but open to other ideas!).
- Planner can type: “Tooth brushs may sell +50% over Christmas—what impact will that have on current stock and what POs do I place?”
- System replies with a file download link and a text summary.
5. Minimal Dashboards
- React page with 5 charts (trend by item code, actual usage vs forecast, safety stock, fill rates, inventory view).
6. Monitoring & Alerts
- Prefect 2 orchestrates tasks and retries.
- Sentry captures stack traces.
- Teams webhook notifies on missing files, schema drift or forecast failures.
Suggested Tech
- Python 3.11, FastAPI, Docker‑compose
- PostgreSQL or SQL Server (on our on‑prem Windows server)
- React + Tailwind front‑end (customer‑facing)
- Streamlit for internal admin only
- Prefect 2 + Sentry for observability
- OpenAI o3 for chat parsing (keys provided), or other LLM if better/cheaper but not going to push out budget or timeline
Deliverables & Milestones
1. Upload & Mapping UI + saved template (Week 2) – 20%
2. Forecast API (Prophet) + nightly files export (Week 4-5) – 30%
3. Chat engine with file output (Week 6) – 25%
4. Monitoring (Prefect 2, Sentry, Teams alerts) (Week 6) – 15%
5. Docker‑compose, README, hand‑over (Week 7) – 10%
Budget & Timeline
- Fixed‑price: US $5 000 – $8 000
- Timeline: 6-7 weeks full‑time
To Bid
- Provide a one‑page architecture sketch (PDF/Figma).
- Share links to ONE forecasting project AND ONE React/Tailwind UI you built.
- Confirm you can start within 7 days of award.
- List your three biggest assumptions or risks.
Only bids inside the budget range and containing all the items above will be reviewed.
There is a lovely pipeline of work following this, and we have a long wishlist, but we'll roll it out as we review this deliverable's success.
Build Lean MVP: Forecast & Inventory Planner (Python + React) — Fixed Price US $7–8k
PROJECT DESCRIPTION
Context
We need a lightweight web tool to replace an Excel‑based forecasting spreadsheet system. The output feeds an existing distribution engine (out of scope here).
Scope – Phase 1 MVP
1. Data Upload & Mapping
- Drag‑and‑drop or SFTP file upload.
- Web UI (React + Tailwind) lets users map columns to target fields; template saved per customer.
- Ingest data to a Postgres or SQL Server database
2. Forecast Service
- Prophet/XGBoost (or similar) produces baseline demand and confidence bands using statistical methods for trend, seasonality, consistency, etc. using the supplied data and inferences from it
- Get data feed from mapped files, establish inventory levels and current movements (inbound, out, committed, etc) for each item
3. Nightly Files Export
- After each nightly run, generate “Binlevels_.csv” containing stock, inbound, outbound, SPL‑Min and shortage per SKU, as well as key forecasting data per item-location combination.
- Generate a suggested purchase orders file (based on item lead times and repair re-work lead times if applicable)
- Generate an overall inventory status file (with item and location attributes, stock levels, usage for past 6 months, and various other available info you'll concatenate)
- Generate a SLA-at-risk nightly file (where we believe we have no stock or are vastly understocked, to be expedited, and are very likely to have an incident based on forecast). Send this as an email with a comment.
- Drop the files to a designated network share or SFTP folder.
4. Portal + Chat What‑If
- User to login with a SSO
- Teams/Web chat (OpenAI o3 key supplied, but open to other ideas!).
- Planner can type: “Tooth brushs may sell +50% over Christmas—what impact will that have on current stock and what POs do I place?”
- System replies with a file download link and a text summary.
5. Minimal Dashboards
- React page with 5 charts (trend by item code, actual usage vs forecast, safety stock, fill rates, inventory view).
6. Monitoring & Alerts
- Prefect 2 orchestrates tasks and retries.
- Sentry captures stack traces.
- Teams webhook notifies on missing files, schema drift or forecast failures.
Suggested Tech
- Python 3.11, FastAPI, Docker‑compose
- PostgreSQL or SQL Server (on our on‑prem Windows server)
- React + Tailwind front‑end (customer‑facing)
- Streamlit for internal admin only
- Prefect 2 + Sentry for observability
- OpenAI o3 for chat parsing (keys provided), or other LLM if better/cheaper but not going to push out budget or timeline
Deliverables & Milestones
1. Upload & Mapping UI + saved template (Week 2) – 20%
2. Forecast API (Prophet) + nightly files export (Week 4-5) – 30%
3. Chat engine with file output (Week 6) – 25%
4. Monitoring (Prefect 2, Sentry, Teams alerts) (Week 6) – 15%
5. Docker‑compose, README, hand‑over (Week 7) – 10%
Budget & Timeline
- Fixed‑price: US $5 000 – $8 000
- Timeline: 6-7 weeks full‑time
To Bid
- Provide a one‑page architecture sketch (PDF/Figma).
- Share links to ONE forecasting project AND ONE React/Tailwind UI you built.
- Confirm you can start within 7 days of award.
- List your three biggest assumptions or risks.
Only bids inside the budget range and containing all the items above will be reviewed.
There is a lovely pipeline of work following this, and we have a long wishlist, but we'll roll it out as we review this deliverable's success.