Courses

PhD coursework, earlier coursework, and teaching. Course notes are in the research diary.

PhD (President's Graduate Fellowship)

Singapore University of Technology and Design, 2026 onwards

  • Multimodal Generative AI (Grade: A)

    How generative models work across text, images and audio: transformers from the ground up, pretraining, and the methods that line up different modalities so one model can take them together.

    transformers · tokenisation · pretraining · vision-language models · contrastive learning · diffusion models · retrieval-augmented generation · multimodal alignment

  • Systems Security (Grade: A) · Project: Humanoid Robot Security

    How attackers and defenders think about whole systems: memory safety, web and network security, the public key infrastructure of the web, authentication, anonymity and side channels.

    memory safety · web PKI · network and DNS security · user authentication · anonymity and privacy · side channels · threat modelling · robot and middleware security

  • Secure Cyber-Physical Systems (Grade: A+)

    Security of industrial control systems: PLC programming, fieldbus protocols, attack models against sensors and actuators, anomaly and invariant based detection, and formal modelling of plant behaviour.

    PLC programming · industrial protocols · attack models · anomaly detection · process invariants · control · timed automata · resilience assessment

  • Foundations of Robotics and Multi-Robot Systems (Grade: A+)

    The building blocks of mobile robots and robot teams: feedback control, motion planning, computer vision and state estimation, and the coordination methods that let several robots work together.

    feedback control · motion planning · computer vision · state estimation · SLAM · consensus and coordination · task assignment · multi-robot path planning

  • Reinforcement Learning for Embodied AI (Audit) · Project: Paddle Duel

    Reinforcement learning for agents that act in physical or simulated worlds: tabular methods, function approximation, deep Q-learning, policy gradients, imitation learning and self-play.

    Markov decision processes · temporal-difference learning · function approximation · deep Q-networks · policy gradients · actor-critic methods · imitation learning · self-play

  • Networked Embodied AI

    Multimodal foundation models for robots and the network underneath them: vision transformers, diffusion and language models, vision-language-action models, robot simulation, and edge to cloud inference under bandwidth and latency limits.

    vision transformers · diffusion models · vision-language-action models · robot simulation · wireless foundations · edge-cloud inference · federated learning · distributed SLAM

  • Robotics Enterprise (Audit)

    The business side of robotics: how a robot moves from a lab prototype to a deployed product, covering markets, operations and what it takes to run a robotics venture.

    robotics commercialisation · product development · deployment · markets and operations · entrepreneurship

  • iDeAL: Innovating with Design and AI for Leaders (Short course, Graduate Professional Development)

    A short programme on design thinking with generative AI: ideation, AI-assisted discovery, multi-agent LLM workflows, and the ethics of building with AI.

    design thinking · generative ideation · multi-agent LLM workflows · AI ethics

B.E. (Hons.) Computer Science

Birla Institute of Technology and Science, Pilani, 2012 - 2016

  • Robotics
  • Computer Networks
  • Pervasive Computing (Department rank 1)
  • Satellite photogrammetry (internship, Indian Space Research Organisation)

Teaching and training

Training Lead, iTrust, SUTD 2018-2026
Curriculum design and the GURU course delivery platform for operational technology security, built with the Ministry of Defence (MINDEF), Singapore, for CISS and CIDEX under Project CARBO.
Judge, Critical Infrastructure Security Showdown (CISS) and CIDEX 2019-2025
Judged cyber-physical exercises alongside experts from MINDEF, CSA, PUB, NUS and Dragos.
Green team and red team infrastructure, NATO CCDCOE Locked Shields and Crossed Swords
Load balancers, scoring, attack launchers, AR/VR visualisers and detectors for NATO CCDCOE's live-fire cyber exercises.