Bharat Yalavarthi

Bharat Yalavarthi

PhD Student

Department of Computer Science
University at Buffalo

I am a second-year PhD student advised by Dr. Nalini Ratha and Dr. Venu Govindaraju. My research focuses on improving interpretability of deep learning models and using interpretability techniques to improve performance, robustness, safety, and fairness of deep learning models.

Research Interests

  • Interpretability
  • Robustness
  • Fairness
  • AI Safety & Alignment
  • Concept Bottleneck Models
  • Multi-Modal Models

Education

  • Ph.D. in Computer Science 2023 – Present
    University at Buffalo, SUNY

    Advisors: Dr. Nalini Ratha & Dr. Venu Govindaraju

  • M.S. in Computer Science 2022 – 2024
    University at Buffalo, SUNY
  • B.Tech. in Computer Science 2016 – 2020
    VIT Unviersity

Publications

  • Label-Free Mitigation of Spurious Correlations in VLMs Using Sparse Autoencoders B.C Yalavarthi, N. Ratha, V. Govindaraju ICLR 2026 - International Conference on Learning Representations
    Paper
  • Shielding Latent Face Representations From Privacy Attacks A. R. Kaushik, B. C. Yalavarthi, A. Ross, V. Boddeti, N. Ratha IEEE FG 2025 - International Conference on Automatic Face and Gesture Recognition
    Paper
  • Aligning Characteristic Descriptors with Images for Human-Expert-like Explainability B. C. Yalavarthi, N. Ratha NeurIPS Workshop 2024 - Interpretable AI Workshop
    Paper
  • Enhancing Privacy in Face Analytics Using Fully Homomorphic Encryption B. Yalavarthi, A. R. Kaushik, A. Ross, V. Boddeti, N. Ratha IEEE FG 2024 - International Conference on Automatic Face and Gesture Recognition
    Paper
  • Efficient Convolution Operator in FHE Using Summed Area Table B. Yalavarthi, C. Jutla, N. Ratha ICPR 2024 - International Conference on Pattern Recognition
    Paper

Experience

  • Researcher Aug 2024 – Jan 2025
    SUNY Research Foundation, Buffalo, NY

    Investigated visual cognitive alignment between LLMs and human perception. Leveraged Sparse Autoencoders for mechanistic interpretability to diagnose failure modes in multimodal LLMs.

  • Research Assistant Jan 2023 – May 2024
    SUNY Research Foundation, Buffalo, NY

    Developed FHE-based privacy solutions for biometric systems, reducing attribute leakage by 35%. Designed kernel decomposition techniques achieving 43% reduction in CNN inference latency.

  • Software Engineer Jul 2020 - Oct 2022
    Harman International

    Developed ADAS and route guidance systems for VW Trucks, reducing SW maintenance costs by 50%. Integrated traffic sign recognition module decreasing incorrect information by 90%.

Curriculum Vitae

Download my full CV for details on my education, experience, skills, and awards.

Download CV (PDF)

Contact

I'm open to research internships, and discussions. Feel free to reach out!

Email: bharatcy2@gmail.com