Now

Robotics research under Prof Zelek @UWaterloo - implementing real-time SOTA visual SLAM algorithms

Where I've worked

Extropic AI (Jan - Dec, 2023) - Research Engineer Intern

Built the core codebase for the simulation of our analog thermodynamic-based AI hardware, laying the groundwork for subsequent R&D modelling and analysis initiatives.
Enhanced the training pipeline efficiency, achieving a 2-fold increase in speed through meticulous memory optimization and bottleneck profiling across 100+ GPU training runs
Created an internal state-of-the-art JAX library for tensor network simulation of quantum circuits on CUDA GPUs

Untether AI(Sept - Dec, 2022) - Compute Kernel Engineer Intern

Developed 3 custom kernels in C++ specialized for rapid neural network inference on the UntetherAI chip architecture.

Institute for Quantum Computing(Jan - Apr, 2022) - Research Intern

Researched a method to effectively realize Bose Hubbard Hamiltonian dynamics through Rydberg atom arrays.
Built infrastructure to run 100+ theoretical simulations under various experimental conditions and constraints.
Developed a python package for efficiently calculating sparse matrix exponentials using the Krylov Method.

Zapata Computing(Jun - Aug, 2021) - Quantum AI Intern

Researched more efficient training methods for quantum generative models.
Developed Orquestra codebase to enable effortless development of QGANs (quantum generative adversarial networks).

May 2023 - Integrated Nvidia GPU compatibility into TensorFlow Quantum for large-scale speedups

Became the largest non-google open source contributor to TensorFlow Quantum
Implemented Bazel build infrastructure and cuQuantum SDK GPU kernels for quantum circuit simulation on heterogenous devices.

April 2022 - Authored article on quantum graph neural networks [article]

Wrote a deep dive into classical graph neural networks, graph convolutions & attention-based graph learning, and QML applied to high energy physics [article]

Feb 2022 - Top 5 QHack 2022 Project [github]

Improved a hybrid quantum graph neural network that was used to solve the particle trajectory reconstruction problem [link]

As part of the Qiskit Advocate Mentorship Program, authored the chapter on QGANs for the newly released quantum machine learning course by IBM's Qiskit team [link]

Under the guidance of a research scientist from AWS Braket, built a hybrid quantum-classical version of the Wasserstein generative adversarial network. [link]
Evaluated learnability of conditional QGANs for sequential data prediction [link]
Investigated heuristics for optimal training of QGANs [link]

Used quantum generative adversarial networks for portfolio optimisation
We built an exploratory project applying qGANs and VQE to solve the mean-variance portfolio problem. This placed our team in the 95th percentile of participants in the QHack quantum machine learning hackathon winning over $4000 in AWS credits.

Explained QGANs for quantum computing enthusiasts.
Derived the cost function and other mathematical properties stated in the original paper

Built a general quantum approximate optimization algorithm for solving weighted cases of the Max-Cut problem

An alternative search engine that we’ve developed & trialed, allowing rural South Africans with a mobile phone to be able to send internet search queries & receive back summarized results all through SMS
Placed top 3 in challenge out of 150+ teams — project built as part of a TKS challenge w/ United Nations.

Addressed customer experience holdbacks by providing an easy-to-implement solution filling data gaps in the Instacart to retailer to CGP pipeline.
Won first place in consulting challenge out of 30+ teams.