About

Starting Semester: Fall 2026
Assigned: No
Location: Atlanta

Google

Client Profile

Google's Global Cloud Logistics organization designs, manages, and optimizes the global physical distribution network that powers Google's technical infrastructure, data centers, and hardware operations worldwide. Operating high-velocity regional distribution centers across the Americas, EMEA, and APAC, our organization leverages advanced industrial engineering, data analytics, warehouse management systems, and material handling technologies to achieve industry-leading fulfillment speed, reliability, and global freight efficiency.

Project Description

PROJECT TITLE: Optimizing End-to-End Site Throughput, Order Cycle Time, and Global Network Capacity in High-Volume Distribution Operations

CLIENT NEED & OPERATIONAL CONTEXT:
Our flagship Southeast Regional Distribution Hub in Fairburn, Georgia operates a complex, high-volume fulfillment environment serving critical demand across North America and international markets. Over the past 12+ months, the facility processed hundreds of thousands of outbound orders and dispatched hundreds of thousands of pallets across outbound trailer loads.

While the facility demonstrates exceptional velocity on a large portion of its volume—fulfilling 25% of all outbound orders in are processed at an extremely high rate and 50% of all orders are processed (P50 median)—a comprehensive 52-week macro throughput analysis revealed severe systemic volatility and cycle-time skewness across three operational dimensions:

1. EXTREME ORDER TURNAROUND SKEWNESS (10.3:1 Tail Ratio):
When comparing the median order turnaround time from order creation to the arithmetic mean is we see the mean is 3.6x the median and the 90th percentile (P90) explodes representing a 10.3-to-1 ratio between P90 tail latency and median performance. Furthermore, weekly median turnaround fluctuates wildly across the year, swinging wildly from balanced weeks to peak congestion weeks.

2. INTRA-WEEK DEMAND & DISPATCH SAWTOOTH IMBALANCE:
Order arrivals exhibit severe mid-week concentration: 64.5% of all weekly order volume drops into the facility between Tuesday and Thursday, while weekend order arrivals drop by over 85%. Conversely, outbound dock dispatches hit a processing ceiling mid-week, creating a recurring Wednesday–Friday work-in-process (WIP) backlog wave that spills into the weekend—forcing Saturday and Sunday shifts to ship nearly 25% of weekly pallet volume just to clear accumulated mid-week backlog.

3. INBOUND VS. OUTBOUND FLOW VOLATILITY:
Weekly outbound load dispatches vary by 2.13x between trough and peak weeks (CV = 0.14), whereas weekly inbound receiving loads fluctuate by 5.43x (CV = 0.23). When high-volume inbound receiving waves collide with mid-week outbound demand surges, internal staging, put-away, and fulfillment workflows experience acute congestion.

PROJECT OBJECTIVES & STUDENT CHALLENGE:
We challenge the Georgia Tech ISyE Senior Design team to ingest 52+ weeks of sanitized empirical transaction data, conduct on-site process mapping at the Fairburn facility, identify the root causes of site throughput constraints, and design an actionable engineering roadmap. Specifically, the team will:

1. Empirical Data Mining & Bottleneck Diagnosis: Analyze 12 months of order-level, shipment-level, and timestamp progression datasets (from inbound receipt/put-away through outbound order creation, allocation, picking, staging, and trailer dispatch) to mathematically isolate where queue delays, batching rules, and flow imbalances generate the 10.3:1 turnaround tail.
2. End-to-End Process Mapping & Capacity Analysis: Map physical and information workflows across the facility to determine the true theoretical throughput ceiling of the site and identify what operational, scheduling, or flow-control changes are needed to close the performance gap.
3. Optimization & Simulation Modeling: Build mathematical optimization and/or discrete-event simulation models to evaluate proposed interventions (such as wave release smoothing, dynamic resource reallocation, staging buffer management, and dock scheduling) to compress P90 order turnaround by >50% and increase peak weekly load shipping capacity by +20% to +30%.
4. Global Footprint Extrapolation & Executive Presentation: Develop a network scaling model that estimates how implementing these throughput improvements across Google's global distribution network impacts enterprise performance—quantifying results in terms of incremental annual global transaction capacity (+15% to +25%+) and Full Truckload (FTL) equivalents eliminated through improved trailer cube utilization, eliminated split shipments, and avoided overflow storage. The team will present their final business case directly to organizational senior leadership.

Skills

- Data Analytics & Exploratory Data Mining: Strong proficiency in SQL and Python/R to query, clean, and analyze large-scale empirical warehouse transaction logs (hundreds of thousands of order and shipment records) and characterize multi-stage lead-time distributions.
- Process Mapping & Industrial Engineering Diagnostics: Ability to conduct on-site gemba walks, Value Stream Mapping (VSM), time-and-motion analysis, and Little's Law / bottleneck throughput calculations across complex warehouse operations.
- Discrete-Event Simulation & Queueing Theory: Experience modeling stochastic arrival/service processes and multi-stage queueing networks (using Simio, Arena, AnyLogic, FlexSim, or Python SimPy) to test capacity ceilings and wave-smoothing policies.
- Mathematical Optimization: Formulating Linear / Mixed-Integer Programming (LP/MILP) models (Gurobi, OR-Tools, PuLP) for workload balancing, dock scheduling, and resource allocation.
- Executive Business Case & Communication: Ability to translate technical industrial engineering models into clear percentile improvements, truckload/transaction equivalents, and executive presentations for Senior Leadership.

Data Access Requirement