Resume

Request CV

Research Engineer at InstaDeep working on reinforcement learning, large language models, and applied AI systems. My background spans self-supervised learning, NLP for African languages, source code understanding, and production software engineering.

Skills

PythonJAXPyTorchTensorFlowHugging Face TransformersReinforcement learningSelf-supervised learningNLPDockerGitLabWeights & BiasesReactReact NativeTypeScript

Experience

  • Research Engineer / InstaDeep

    2024 - Present
    • Working in research on reinforcement learning, large language models, and AI systems.
    • Recent research includes inference strategies for complex reinforcement learning and sequence models for offline multi-agent reinforcement learning.

  • Engineering Intern / InstaDeep, DeepPack team

    Aug 2023 - Mar 2024
    • Applied quality diversity and evolutionary computing methods to speed up model training in a production-grade reinforcement learning system.
    • Fixed a continuous deployment issue affecting the machine learning experiment workflow for a team of engineers.
    • Used Python, JAX, TensorFlow, GitLab, Docker, Pandas, and Matplotlib.

  • Software Engineering Intern / Boostasoft

    Feb 2020 - Feb 2021
    • Led front-end development for a production recruitment platform serving over 1,000 users.
    • Managed development of an Android COVID-19 tracking app with real-time statistics across more than 20 countries.
    • Used JavaScript, TypeScript, React, React Native, and GitLab.

Selected Projects

  • LLM fine-tuning on African languages / Personal research project

    Nov 2021 - Dec 2023
    • Studied how NER annotation density affects pre-trained language models in low-resource African language settings.
    • Built training and evaluation notebooks for BERT, RoBERTa, and multilingual BERT variants across Pidgin, Kinyarwanda, and Swahili.
    • Published the resulting token-classification model cards on Hugging Face.

  • MiLMo - Minecraft Language Model / Personal project

    Aug 2022 - Jul 2023
    • Trained a small GPT-style model on Minecraft maps encoded as 3D NumPy arrays.
    • Built scripts to extract map data from a running Minecraft server, generate new structures, and visualize model outputs back in Minecraft.

  • Entelect 2023 RL Game Bot / Personal project

    Apr 2023 - Aug 2023
    • Built an AI bot for the Entelect Challenge 2023 with the Wits Applied AI Interest Group.
    • Set up a reinforcement learning training pipeline around the CyFi game server, including PPO experiments and environment tooling.
    • Used Python, C#, PyTorch, Docker, Hydra, and Weights & Biases.

Education

  • M.Sc. Computer Science / University of the Witwatersrand

    2022 - 2024
    • Research topic: a novel data augmentation method for contrastive self-supervised learning.
    • Supported by a Google DeepMind Scholarship.

  • B.Sc. (Hons) Computer Science, cum laude / University of the Witwatersrand

    2021
    • Research topic: influence of contrastive learning on source code plagiarism detection.
    • Received Dean's List Honours and a Postgraduate Merit Award.

  • B.Sc. Applied Physics / University of Buea

    2016 - 2019
    • Final-year research topic: introduction to quantum computing, theory and applications.

Selected Publications

Awards & Certificates

  • Google DeepMind Scholarship / University of the Witwatersrand

    2022

  • Runner-up Award / Minecraft Open-Endedness Challenge, GECCO

    2022

  • Dean's List Honours Award / University of the Witwatersrand

    2021

  • Postgraduate Merit Award / University of the Witwatersrand

    2021

  • TensorFlow Developer Certificate / DeepLearning.AI / Coursera

    2021