Project Almanac - Optimized Timetable Scheduling at Lower Costs
www.projectalmanac.org
Watch the demonstration video: • Project Almanac - Demonstration
In this video, I discuss a project that I've been working on, titled Project Almanac.
Project Almanac offers a robust, flexible architecture that doesn't try to build a perfect result in one go. Instead, we use an iterative approach powered by Artificial Intelligence and Machine Learning. We feed "soft constraints"—like staff salaries, facility waste, and student exhaustion—into a custom Neural Network. Our system predicts the most optimal event swaps to reduce "heat distribution" (the cost gap between slots), generating a highly optimized, "good enough" timetable far faster than traditional methods.
The true competitive advantage of Project Almanac lies in its hardware-software synergy. Advanced optimization is resource-heavy, but we’ve eliminated the need for expensive servers. We utilize open-source cluster computing to distribute workloads across multiple nodes. Our testing proved the system can run effectively on discarded, "older generation" desktops and laptops using the Linux-based Message-Passing Interface (MPI). Through this parallel processing, we’ve achieved a quantitative reduction in complexity—moving from O(n6) in brute-force methods to a linear O(n) with our AI optimization.
Project Almanac is a modular, scalable proof of concept ready for industry application. The architecture allows institutions to scale their hardware up or down based on their specific needs and budget. While we are starting with educational institutions, this same AI-driven cluster architecture can be generalized for train schedules, product releases, complex trading schedules, route handling, etc. as well.