Hi, I'm Andrew Gordienko 👋
I build solutions for real-world problems by following the data. That means collecting messy inputs, validating them, and turning them into systems people use, from infrastructure to applied AI.
A real-time power outage map and API for Canada. Aggregation, normalization, and delivery of outage data with a focus on reliability and speed.
We help teams modernize legacy systems through carve-outs, integration layers, data tooling, and testing support alongside larger programs.
A research and tooling effort to find gaps in government policy and execution, starting with construction delays, using LLMs to surface patterns and root causes.
A simulation and optimization project to forecast station shortages and generate rebalance plans, including event-driven demand signals.
A racing sim playground to explore vehicle dynamics, overtakes, track geometry, and performance engineering workflows.
A chess engine and interactive sandbox you can play with in-browser. Lightweight, fast iteration, and a clean analysis UI.
A hybrid locomotion pipeline that bootstraps walking from motion-captured human joint angles using NEAT, then fine-tunes with DDPG for generalization in rough terrain. The main bottleneck was observation design.
Research at UTMIST (University of Toronto Machine Intelligence Student Team) on evolving creature morphologies, training a universal PPO walking policy, and coordinating 2v2 soccer using grid-based planning with AlphaZero-style fine-tuning.