Ilia SMIRNOV
Resume: HTML | pdf
Hello and welcome to Ilia's personal web page!
I am currently working as a programmer post-doc at the Intelligent Transportation Systems lab at the University of Toronto.

The chief purpose of this page is to document personal projects I undertake to increase my programming skill and knowledge. My current main languages are C/C++ (with CUDA) and Python, as well as Javascript for the web. Professionally, I am especially interested in Machine Learning, Optimization, Traffic Control, and Computer-Aided Physics Simulation.

Programming projects

Essays
Sep. 2020 Learned image classifiers
kNN, Logistic regression, Fully-connected NNs, Convolutional NNs in CUDA C++
Jul. 2020 Q-learning and Tic-Tac-Toe
Tabular Q-learning and DQN, as well as Minimax and another Rules-Based agent, all in C++
Shorter notes
Nov. 2020 Mask R-CNN - based filter
Made for a motion-triggered garden camera. Filters out captures of animals and humans from false positives of trees swaying in the wind.
Nov. 2020 CUDA vs. Tensor cores
Three methods (naive CUDA, WMMA, cuBLAS) compared for multiplying matrices and for training fully-connected neural nets
Oct. 2020 Implementing modern machine learning methods in PyTorch
ResNets completed, working on Transformers, GANs, VAEs

Mathematical interests

Algebraic Geometry and connected areas (Number Theory, Representation Theory, Complex Analysis). Within Algebraic Geometry, I am especially interested in Enumerative Geometry approached through Intersection Theory.