Supply chain & applied AI
14 public repositories, each starting from an operating problem and ending in a decision a planner can act on. Every one is scored against the baseline it has to beat, reports what didn’t win, and ships a short walkthrough video and an audit report.
Which model to trust per SKU, and forecasts judged by the stock they let you hold.
Half the catalogue sells nothing most weeks, so one method and MAPE pick the wrong model: segment-wise backtest of seven forecasters.
LightGBM cuts eight-week WAPE to 21.9% from 29.1% for seasonal naive
Safety stock covers what a smoothed order history misses: SQL-joined customer × SKU panel and two-stage boosted forecasts judged on holding cost.
Holding cost at 98% fill falls 48.7% against per-line ETS
Buffers, policies and capacity plans tested against the same demand before anything changes.
A distributor holds 172 days of stock and still stocks out: DDMRP buffers against min-max and reorder-point policies at matched fill rate.
DDMRP holds 52.1 days of inventory against 97.6 for legacy min-max (−47%)
Policy changes cost a quarter of empty shelves to try for real: a SimPy twin of 2 DCs and 36 stores that accepts a change only if it beats its own error.
Ordering every two days alone: -142,008 ±3,375 per quarter (about 6%)
Riders must be committed weeks before Ramadan and a 4.2x mega-sale: event-aware forecast, MIP shift plan and an order-by-order replay.
17.6% less rider cost, 6.4% less total cost including lost margin
Vehicle routing, dispatch and the experiment that says what actually moved.
A 300-truck fleet misroutes ~4% of shipments and books drivers a week ahead from a forecast: time-windowed VRP routing with OR-Tools plus a zone-demand forecast.
Peak week: -22.1% road km against the sweep dispatcher (140,683 to 109,607)
Quick-commerce hubs dispatching one order per rider by hand miss the 30-minute promise: order bagging, Hungarian rider matching and a hub-randomised A/B test.
Fleet cost per delivered order 9.79 to 8.75 AED (-10.6%)
Markdowns, order margin, targeting and next-best-action, each priced in money.
A blanket 30% clearance markdown gives margin away on some SKUs and writes off stock on others: per-SKU elasticity and markdown-path search.
Promotional margin 56,628 against 18,877 for the 30% rule
One order in five ends in a return, cancellation, refusal or chargeback: a five-outcome model priced with unit economics, and a COD switch-off rule.
Sale-week daily margin error: 4.3% against 22.5% for finance's flat rates
Prescription deciles barely separate launch converters (AUC 0.623): referral-network features and embeddings in a paired ablation.
Graph features lift AUC from 0.641 to 0.668 (+0.027, 95% CI +0.018 to +0.037)
A commercial team must pick content, channel and message for 20,000 prescribers using approved claims only: sequence recommender, channel policy and governed RAG.
206.6 engaged per 1,000 touches against 189.8 logged (+8.9%)
LLM cost control and agentic data engineering, scored against a known answer.
Every LLM request goes to the 7B model because nobody knows when a 3B answer is safe: a verified small-to-large cascade.
82.5% completion against 85.0% for the large model alone, at $0.0176 per 1k requests against $0.0452 (61% cheaper)
Onboarding a new source takes days of manual type, key and join detective work: agents that infer them and emit a tested dbt project and Airflow DAG.
88.6% of 45,677 injected defects caught at 98.4% row precision
Maintenance scheduled by expected savings under a fixed crew.
A crew can do about six planned jobs a week across 200 assets: failure-risk model, break-even rule and crew-budgeted scheduler run closed-loop.
unplanned failures 44.2 vs 111.2 a year, cost 2.02M vs 4.01M a year