harrywaugh Goto Github PK

activitytrackingwithsensorfusion

I wrote my dissertation for my Computer Science degree on whether the accuracy of GPS tracking for sports activities (running, cycling, walking) could be improved using additional sensors found in a smartphone. Initially, I implemented a Kalman Filter to fuse the readings of each sensor, this was written in python and used popular libraries like NumPy. I then researched and taught myself the theory behind Recurrent Neural Networks (RNNs) in order to implement them using tensorflow and keras. This project was supported by the Oracle Innovation Accelerator program, who generously gave me access to a cloud compute instance.

amgx

Distributed multigrid linear solver library on GPU

blas_tracing_tests

Some simple tests that can be run with the Arm tracing tool, to check that calls to the BLAS library are correct

build_mgcfd

c-term-1-year-1-

coms30115

cv

environment

first-contributions

🚀✨ Help beginners contribute to open source projects

hpc-course-getting-started

A tutorial-style introduction to the HPC course

hw16471.github.io

Visit my page!

markdown-cheatsheet

Markdown Cheatsheet for Github Readme.md

pbmpi

phylobayes mpi

perf-libs-tools

scotlandyard

superscalarsim

Advanced Architecture Coursework at the University of Bristol. Superscalar simulator.

uob_appliedsecurity

This project developed my understanding of how AES encryption works and the different ‘trade offs’ between speed, memory usage and ultimately, security. An AES implementation was first developed to run a scale-board from the University of Bristol, this device has severe hardware limitations and thus any implementation must be efficient. One of the ways this was done was to pre-compute the sbox function. A Differential Power Attack (DPA) was then created in python to interact with the board in order to gather power traces. These recorded traces could then be compared to hypothetical power consumption values for a chosen message in order to recover the key, using approximately 125 traces. A counter-measure for this style of attack was implemented called masking, this carefully used 6 masks throughout the encryption process to prevent information leaking from the CPU. This prevented key recovery for ‘first order’ DPA style attacks.

uob_clustering_and_classifying_data

uob_gameoflife

uob_operating_system_kernel

uob_raytracer

A working raytracer written in C++ using the GLM and SDL libraries.

uob_web_tech_cw

This project involved creating a website for a University of Bristol games project, Khaos Kitchen. The website showcased a range of different technologies, each used with respect to web standards and compatability. A working forum was set up using a NodeJS server and an SQLite3 embedded database. The forum was created dynamically using templating engine ‘pug.js’, and requested different parts of the forum ‘on the fly’ as the user needed them.

virtual-assistant

A linux based Virtual assistant on Artificial Intelligence in C