We are looking for a strong Analytics Engineer / BI Developer to help build a lightweight food waste and buying decision pilot for a coffee shop with multiple branches.
This is not a heavy app build.
It is a fast, practical pilot focused on turning existing reports into a simple dashboard, rules-based recommendations, and automated weekly recommendation / action packs.
The goal is to help the business:
- reduce food waste
-improve buying decisions
- identify products to keep / reduce / review
- highlight branches needing intervention
- generate a repeatable weekly operating process
Current situation
The business already has exported reports covering:
product sales
inventory / control / purchased vs sold vs waste
manual waste inputs
Some waste data is manual, so part of the pilot will involve creating a clean structured waste input method and a master mapping table so product and branch data align properly.
What needs to be built
We need someone who can help us build the pilot end-to-end, including:
1. Data model / setup
- Create a product master / SKU mapping table
- Create a branch master table
- Standardise data from multiple reports
- Handle messy real-world Excel / CSV inputs
2. Data ingestion / transformation
- Import weekly Excel / CSV reports
- Clean and map products and branches
Calculate core metrics such as:
- purchased qty
- sold qty
- waste qty
- waste %
- sell-through
- trend vs previous period
- branch risk score
- recommendation status
3. Recommendation logic
Build a simple rules-based recommendation engine to classify products / branches into:
- Keep
- Reduce
- Review / Remove
- Monitor
Each recommendation should include a short rationale, for example:
- zero sales with recorded waste
- overbought vs demand
- high waste relative to sell-through
- weak in this branch vs stronger stores
4. Dashboard
Build a simple internal dashboard with key views such as:
- Executive Overview
- Branch Performance
- Product Performance
- Action Centre
Missing / late waste submissions (optional but preferred)
The dashboard should be practical and ops-focused, not overly designed.
5. Manual waste capture
Help implement a lightweight manual waste input method, likely via:
Google Form, or simple web form, or controlled upload template
We do not want a heavy branch-facing app for Phase 1.
6. Automated weekly output
Set up a weekly automated action pack / email process:
- owner / ops summary email
- optional branch-level email
- PDF / HTML summary or dashboard link
Tech approach
We are open to the best practical stack, but likely something like:
Python + pandas
SQL / SQLite / Postgres
Power BI / Metabase / Looker Studio
lightweight email automation via Python / n8n / Zapier / Make
We want the fastest credible solution, not overengineering.
Important constraints
This is a pilot, so we are not looking for:
- full custom web app
- ML forecasting
- mobile app
- order automation
- deep integrations / APIs in Phase 1
What success looks like
At the end of the pilot, we want:
- clean mapped data across reports
- a working dashboard
- product / branch recommendations updating from the data
- weekly action email output
- a setup that can be refreshed each week with minimal manual effort
Ideal profile
We are looking for someone who:
- is strong in Python / pandas / SQL
- is comfortable with messy Excel / CSV operational data
- can build practical dashboards quickly
- understands how to translate business rules into decision logic
- can work fast and communicate clearly
ideally has experience in retail, F&B, inventory, ops analytics, or BI
Nice to have
- experience with branch/store performance analysis
- experience with food retail / hospitality / stock control
- experience with lightweight automation workflows
To apply COMMENT 'COFFEE IS LIFE' to show you read the brief
Please include:
A short note on similar work you’ve done
Your preferred stack for this type of pilot
Examples of dashboards / BI work you’ve built
How you would approach messy SKU / branch mapping
Your estimated timeline for a first usable version
Project style
We are looking for someone practical, commercial, and fast.
This should feel like a working pilot in a real business, not an overbuilt data project.
Apply Now
Apply Now