Bachina Pranav

prof_pic.jpg

Hi, I’m Bachina Pranav. I am a Machine Learning Engineer at FactSet Research Systems. I graduated from the International Institute of Information Technology, Hyderabad with a B.Tech (Hons) in Computer Science.

During my academic journey, I worked closely as an undergraduate researcher under the supervision of Dr. Charu Sharma and Dr. Ankit Gangwal.

My primary research interests revolve around Machine Learning Security, Geometric Deep Learning, and AI Safety.

For students interested in entering the field, I created A Reading Path into Machine Learning Security Research, a structured self-study guide covering foundational concepts, seminal papers, and hands-on research checkpoints.

selected publications

  1. genie.png
    Genie: Watermarking graph neural networks for link prediction
    Venkata Sai Pranav Bachina, Ankit Gangwal, Aaryan Ajay Sharma, and 1 more author
    Transactions on Machine Learning Research, 2026
  2. loretta.png
    LORETTA: A Low Resource Framework to Poison Continuous Time Dynamic Graphs
    Himanshu Pal, Venkata Sai Pranav Bachina, Ankit Gangwal, and 1 more author
    In Proceedings of the AAAI Conference on Artificial Intelligence, 2026

news

Jun 2026 Presented our paper “PatchPoison: Poisoning Multi-View Datasets to Degrade 3D Reconstruction” at CVPR 2026 in Denver, USA.
Apr 2026 Our paper “PatchPoison: Poisoning Multi-View Datasets to Degrade 3D Reconstruction” was accepted at SPAR-3D Workshop at CVPR 2026.
Mar 2026 Our paper “Genie: Watermarking Graph Neural Networks for Link Prediction” was accepted at TMLR.
Jan 2026 Presented our paper “LORETTA:A Low Resource Framework to Poison Continuous Time Dynamic Graphs” at AAAI 2026 in Singapore.
Nov 2025 Our paper “LORETTA:A Low Resource Framework to Poison Continuous Time Dynamic Graphs” was accepted at AAAI 2026.
Aug 2025 We presented our paper “Dissappearing Ink: How Partial Model Extraction erases Watermarks” at AsiaCCS 2025 in Hanoi, Vietnam.
May 2025 Starting working full-time as a Machine Learning Engineer at Factset Research Systems
Apr 2025 Our paper “Dissappearing Ink: How Partial Model Extraction erases Watermarks” was accepted at AsiaCCS 2025.