Problem Statement

Manually calculating and continually updating Estimated Times of Arrival (ETAs) for trucks, and subsequently coordinating and scheduling appointments at logistics facilities within various Transportation Management System (TMS) portals, is an incredibly time-consuming, error-prone, and reactive process. This manual overhead leads to significant operational inefficiencies, increased truck detention costs, suboptimal facility utilization, and a lack of real-time visibility, causing frustration for dispatchers, carriers, and customers alike.

Goal

The primary goal of this project was to design and implement an automated system for ETA calculation, prediction, and appointment management directly integrated with TMS portals. This system aims to provide real-time, accurate ETAs, intelligently optimize and schedule appointments, and automate communication, thereby dramatically improving operational efficiency, reducing costs associated with delays, and enhancing transparency across the logistics supply chain.

Tech Stack

Python, FastAPI, LangChain, LangGraph, LangSmith, Azure Cognitive Services (GPT 4.1), PyTorch, Pandas, SQL, Kafka, MongoDB, PostgreSQL, Playwright, Selenium

Impact & Opportunity

• Significant Operational Efficiency Gains: Reduced manual effort in ETA tracking and appointment scheduling by up to 85%, allowing dispatchers and logistics coordinators to manage more loads with less overhead.
• Reduced Detention and Waiting Times: Optimized scheduling led to a 20-30% decrease in truck detention times at facilities, resulting in cost savings and improved carrier satisfaction.
• Enhanced Customer and Carrier Satisfaction: Provided real-time visibility and accurate predictability, leading to improved communication, reduced frustration, and stronger partnerships.
• Improved Resource Utilization: Maximized the efficiency of trucks, drivers, and facility docks through intelligent, dynamic scheduling.
• Scalability and Adaptability: Developed a flexible solution capable of integrating with multiple TMS platforms and adapting to evolving logistics requirements.

Key Contributions & Architecture

• Designed and implemented algorithms to accurately calculate and predict truck ETAs by integrating various data sources (e.g., GPS tracking, traffic conditions, driver hours-of-service, historical performance).
• Leveraged machine learning models (e.g., time-series forecasting, regression) to improve ETA accuracy, accounting for variables like weather, route specifics, and potential delays.
• Ensured dynamic updates to ETAs based on real-time events and data feeds.
• Developed an intelligent scheduling engine that automatically proposes optimal appointment slots for pickups and deliveries based on facility availability, truck capacity, driver schedules, and real-time ETAs.
• Implemented logic for slot optimization, minimizing wait times and maximizing facility throughput.
• Supported dynamic rescheduling capabilities, automatically adjusting appointments in response to updated ETAs or unforeseen disruptions.
• Established robust API and custom portal integrations with various TMS (Transportation Management System) portals (specific vendor APIs, custom enterprise systems) to pull and push ETA and appointment data.
• Automated the entire workflow from initial load assignment to final delivery, including: Automatic submission of appointment requests; real-time status updates and notifications within the TMS; handling of appointment confirmations, rejections, and rescheduling requests.
• Implemented automated notification systems to alert relevant stakeholders (e.g., dispatchers, carriers, customers, facility staff) about ETA changes, appointment confirmations, and potential delays via preferred channels (e.g., email, SMS, in-portal alerts).