Home Projects Academic

Academic Projects
@ Cambridge

Thesis and coursework projects during my graduate studies at the University of Cambridge.
If the code/report is not provided and you would like access, please contact me.

Card image cap
Card image cap

Fast Geospatial Data Processing
using Data Compression

MRes thesis on beating main-memory bandwidths for geospatial computation via cache-optimised in-memory data compression. Implemented in C++ with AVX512.
  • Deployed state-of-the-art SIMD-optimised data compression algorithms to reduce the memory footprint and improve cache efficiency during geospatial data processing.
  • Leveraged the AVX512 instruction set in C++ to parallelise compression algorithms, improving performance by maintaining compressed data closer to the CPU.
  • Achieved significant processing speedups, up to 78%, by fusing decompression and aggregation stages, streamlining the handling of large geospatial datasets.
  • Utilised SLURM for managing HPC resources, enabling efficient scaling and execution of 8000 hours of benchmarks and optimisations.
  • Applied multi-objective Pareto optimisation to identify and implement the most effective compression codecs, striking a balance between computational speed and memory efficiency.

    • C++   Python   SLURM   GDAL
    Card image cap

    Embedded Real-Time Activity Recognition

    Embedded Systems for the Internet of Things coursework. Implemented an algorithm in C on a lightweight micro-controller and IMU to quantify the probability of walking activities.
  • Developed in C on a FRDM KL03 micro-controller with an MMA8451Q accelerometer, optimising for space by stripping excess boot code and implementing real-time parameter updates.
  • My algorithm successfully differentiated between walking, jogging, and running based on stride duration probabilities, though it needs tuning to better distinguish jogging and running.
  • Utilised Gaussian distributions for stride timing with dynamic variance updates via Welford's algorithm and efficient CDF calculations using the Abramowitz and Stegun approximation.

    • C
    Card image cap

    Temperature and Pulse Sensors

    Sensor Design Project. Implemented and calibrated a temperature and pulse sensor. Raw electronics implemented on an Arduino, real-time signal processing in C, data analysis in Python.
  • Employed Arduino for real-time processing in C and data analysis in Python, using thermistors and photoplethysmography (PPG) for enhanced sensor functionality.
  • Achieved a 43.2% increase in temperature measurement accuracy through calibration of thermistors.
  • Implemented a custom real-time heart rate detection algorithm using peak-finding techniques to provide immediate user feedback.

    • C   Arduino   Python
    Card image cap

    Bayesian Optimisation for PPO

    Machine Learning and the Physical World coursework. Led a group project using Python to automatically tune PPO's hyperparameters using Bayesian Optimisation.
  • Automatically tuned the hyperparameters for Proximal Policy Optimization (PPO) for the Cartpole problem using Bayesian Optimisation, demonstrating more efficient convergence over random search.
  • Developed in Python using Gymnasium for simulation, Stable Baselines3 for PPO, and GPyOpt for Bayesian Optimisation, focusing on critical hyperparameters like the learning rate and entropy coefficient.
  • Employed Gaussian Process surrogates with Matern52 and periodic kernels, achieving optimal performance through the Expected Improvement acquisition function.
  • Varied neural network architectures to measure the importance of PPO's actor and critic.

    • Python
    Card image cap

    Probabilistic Ranking & Latent Dirichlet Allocation Model

    Probabilistic Machine Learning courseworks. One on TrueSkill / probabilistic ranking, the other on topic modelling / the LDA model.
  • Understood and implemented advanced statistical algorithms in MATLAB.
  • Evaluated convergence of the collapsed Gibbs sampler and LDA models.
  • Achieved a grade A- for both courseworks.

    • MATLAB