Optimization, Search and Reasoning for AI Systems

<p>Long before large language models, AI systems solved real planning, routing, and scheduling problems with a different toolkit entirely: graph search, gradient-based optimization, constraint satisfaction, and adversarial game search. This project follows Idris Okafor, a backend engineer at a fictional last-mile delivery startup called Fernbridge Logistics, as he rebuilds the company's broken dispatch system from the ground up — implementing a real A* pathfinder, tuning a convex cost model with gradient descent, writing a backtracking constraint solver for driver and vehicle assignment, and modeling a competitive loading-dock negotiation with minimax search. Every algorithm is presented through the actual problem it solves, with full, runnable Python code.</p>

Tools: Convex Optimization, A* Search, Heuristic Search, Constraint Satisfaction, Game Search (Minimax)