Lynn Kandakji

Applied AI/ML | UCL PhD

London, England, United Kingdom

About

PhD in Applied Stats and ML. Building probabilistic models and autonomous systems. Code + experiments: github.com/lkandakji

Experience

  • Senior Applied AI/ML Researcher at JPMorganChase
    Dec 2025 - Present · 7 mos

    Chief Analytics Office within the Commercial and Investment Bank, focused on the design and deployment of agentic systems, LLM evaluation and safety, and the end-to-end delivery of high-priority AI products.

  • UCL (London Area, United Kingdom)
    • Postgraduate Teaching Assistant
      Feb 2023 - Jun 2026 · 3 yrs 5 mos

      MSc programmes spanning Bioscience Entrepreneurship and Venture Strategy for Emerging Technology, Foundations of AI, and AI in Healthcare

    • Doctoral Researcher
      Oct 2022 - Dec 2025 · 3 yrs 3 mos

      PhD focused on extracting signal from high-dimensional, low-SNR data using probabilistic modelling, statistical inference, geometric learning, and 3D transformer architectures.

  • Machine Learning Research Fellow at Moorfields Eye Hospital, London
    Nov 2022 - Dec 2025 · 3 yrs 2 mos

    Led ML and deep learning research at a leading global ophthalmic institution, applying predictive models to 3M+ clinical records with deployment potential across tens of thousands of patients, resulting in seven peer-reviewed publications and a six-figure AWS research partnership.

  • Applied AI/ML Researcher at JPMorganChase
    Sep 2024 - May 2025 · 9 mos

    Within Securities Services, led the design and delivery of the firm's first client-facing agentic AI system and a multi-agent post-trade processing pipeline. I embedded AI into core workflows to modernize custody, fund accounting, and trading operations by automating complex financial processes, extract insights from massive datasets, and reduce manual workload at scale across a platform serving 7,000+ institutional investors globally.

  • Machine Learning Intern at CRTC
    May 2022 - Oct 2022 · 6 mos

    Canada’s federal telecommunications regulator, designed and deployed a national-scale fraud detection system processing millions of daily telecom records.