Shreyansh Modi

Projects

Reproducibility studies, research prototypes and things I’ve built.

• RE-UltraBreak: Universal Jailbreak Attacks on VLMs

  A reproducibility study of UltraBreak: universal, cross-model image jailbreaks against black-box VLMs, using randomised image transformations and loss relaxations to maximise transferability.

  Jun 2026 – Aug 2026 · Adversarial ML, VLM security

• VLM Hallucination Mitigation via Self-Verification Decoding

  A decoding-time framework that enforces strict textual grounding to reduce visual hallucinations, validated on the POPE and CHAIR benchmarks.

  Jun 2025 – Sep 2025 · VLMs, decoding

• Reproducibility Study: Diffusion Models Beat GANs

  Replicated OpenAI’s unconditional U-Net and noise-robust classifier to implement classifier guidance, and validated that scaling classifier gradients trades sample diversity for visual fidelity.

  Aug 2025 – Sep 2025 · Diffusion models

  Code

• Adobe Photo Editor

  Context-aware background removal, style transfer and typographic editing, with MobileSAM optimised through ONNX Runtime and INT8 quantisation for real-time edge deployment.

  4th place · Inter IIT Tech Meet 14.0

  Nov 2025 – Dec 2025 · Computer vision

  Code

• InVisionDX: CNN-Based Disease Detection

  A web app that detects COVID-19, pneumonia, tuberculosis, lung cancer and Alzheimer’s disease from medical images with ResNet and VGG16, at over 95% classification accuracy.

  Mar 2025 – Apr 2025 · PyTorch, Flask

  Code

• HireSense

  An open-source NLP tool that uses fine-tuned named-entity recognition to parse resumes and match candidates to job descriptions.

  Python, NER, NLP

  Code