Shipping Intelligence Intern
Shipping Intelligence Intern
About the role
Every order we ship goes through a rate-shopping engine that pulls quotes from UPS and FedEx in parallel, scores them on cost plus historical performance plus invoice variance, and picks a winner. Then it learns from the carrier invoice when it arrives a month later. You will work on the system that turns "every shipment is a small ML training example" into a real margin lever.
What you'll build
- A parser for carrier invoice CSV/EDI files that reconcile billed cost vs quoted cost per tracking number.
- Features for our gradient-boosted regressor that predicts final invoiced cost from package + route + carrier inputs.
- Purchase shipping labels electronically thorough carrier APIs , post tracking number and status on the relevant storefront, group them in batch with same SKU and attach a packaging label to assist Warehouse workers identify the Products
- A finance dashboard showing carrier-level spend, on-time %, damage rate, and invoice variance — useful in annual carrier negotiations.
What you'll learn
- How a real ML training loop works in production: feature stores, holdout evaluation, gated promotion, monitoring.
- Working with carrier APIs (UPS REST, FedEx Web Services) auth, rating, label generation.
- Time-lagged supervised learning (the model gets ground truth a month later it is an interesting problem).
- Python, scikit-learn / XGBoost / LightGBM, Postgres, and S3.
Required
- BS/MS in CS, Data Science, or related.
- Solid Python and a course (or self-study) in ML or statistics.
- Comfortable reading documentation for unfamiliar APIs.
Nice to have
- Curiosity about A/B testing, model monitoring, or ML ops.