Allan Reyes

Project Artemis webdev gt rails ruby

A web application that helps CDC Utah streamline, identify, and correct data discrepancies with newborn screening data.

BGP prefix hijack infosec gt

Configured a virtualized Border Gateway Protocol (BGP) topology, conducted a prefix hijack, and created security protocols to detect and prevent them.

Buffer bloat and congestion control infosec gt

Conducted a comparative analysis on TCP CUBIC and TCP Reno for congestion control and buffer bloat in virtualized networks.

Reproducing Raicu 2012 infosec gt

Reproduced and analyzed Multipath TCP (MPTCP) and experimental results from How Hard Can it Be? Designing and Implementing a Deployable Multipath TCP (Raicu 2012) on a virtualized network.

TCP Fast Open infosec gt

Reproduced and analyzed TCP Fast Open (TFO) on production web servers, including Hacker News, Reddit, and gnu.org.

Project Disco webdev gt

An alternative video format that reduces file size, enables open-sourced maintenance and collaboration, and promotes native translation; presented at Learning with MOOCS 2017 in Austin, TX.

Bipedal walker ml rl

Implemented a reinforcement learning agent using Deep Deterministic Policy Gradients (DDPG) to learn how to control a bipedal walker (OpenAI).

Reproducing Greenwald 2003 ml gt rl

A reproduction of experimental results of Correlated Q-learning (Greenwald 2003).

Lunar lander ml gt rl

A reinforcement learning agent using Deep Q-networks (DQN) to control a lunar lander (OpenAI).

Reproducing Sutton 1988 ml gt rl

A reproduction of experimental results of Learning to Predict by Methods of Temporal Difference Learning (Sutton 1988).

Polymorphic blending attack infosec gt

Implemented a polymorphic blend attack exploit to defeat a payload-based intrusion detection system (PAYL).

Unsupervised learning survey ml

Application and optimization of unsupervised learning and dimensionality reduction algorithms (k-means clustering, expectation-maximization, PCA, ICA, and randomized projections) across the same two datasets from the previous supervised learning survey.

Randomized optimization survey ml

Application and tuning of randomized optimization algorithms (randomized hill-climbing, simulated annealing, genetic algorithms, and MIMIC) on four classical problems (Traveling Salesman, Four Peaks, Knapsack, and ANN optimization).