1. L—01

    Research & benchmarksMar 2026Independent research paper — Understanding, Failure Modes, and Solutions

    GANs Under Data Scarcity

    Controlled study of how four GAN loss functions behave as training data is starved from 25,000 to 1,000 images.

    16Training runs

  2. L—02

    Generative modelsApr 2025Project Trainee — DRDO, Jodhpur

    Painting Generation with CycleGAN

    Unpaired image-to-image translation that turns real photographs into landscape paintings.

    58.4 → 49.6FID

  1. L—03Computer visionEntries in preparation
  2. L—04LLMs & agentic systemsEntries in preparation

More notebooks are being written up — these areas land next.

ESC