Naresh Kumar Devulapally#
I am a PhD candidate in Computer Science and Engineering at the University at Buffalo, SUNY, advised by Dr. Vishnu Lokhande. I am currently a Research Scientist Intern at Adobe Research.
Currently on the job market
I am actively looking for postdoctoral and research scientist positions starting February 2027. Please email me at devulapa@buffalo.edu or devulapally.phd@gmail.com.
- Image/Video Generation and Editing
- Content Provenance/Watermarking
- Hallucination Detection and Mitigation
- Model Interpretability
- Vision-Language-Action/World Models
- Multi-Modal Learning
Research Focus
My research focuses on controlling diffusion models during generation to address challenges in:
AI Provenance and AI Safety
Explain Model Representations
Improve Reliability
My work spans images, video, and language, including text-to-image diffusion models, diffusion language models, and diffusion vision-language models.
Establish provenance for generated media and protect content against unauthorized personalization.
Uncover representations underlying a model’s outputs and turn them into textual explanations.
Detect and correct hallucinations during generation in diffusion models (image and text generation).
Previously, I completed my MS in Computer Science at the University at Buffalo, working on robust multimodal learning for video emotion recognition with Dr. Junsong Yuan and Dr. Sreyasee Das Bhattacharjee, published at ACM Multimedia 2023, BigMM 2023 and ICME 2024. Before graduate school I spent two and a half years as a Senior Data Scientist, building and deploying production machine learning systems. I studied Mechanical Engineering at NIT Tiruchirappalli.
Collaborators#
Publications
Teaching#
CSE Graduate Teaching Award, 2025
Received the CSE Graduate Teaching Award at the University at Buffalo for CSE 4/573: Computer Vision and Image Processing, the first course at UB to include a semester-long module on diffusion models.
News#
-
👨🏻🏫 Invited Guest Lectures on Diffusion Models in CSE 555: Pattern Recognition course at UB.
- Instructor: Prof. Ifeoma Nwogu.
- 🎉 Token-to-Token Similarity Reduction in dLLMs for Edit-Friendly Hard Prompt Inversion is accepted to CVPR 2026!
-
👨🏻🏫 Invited Guest Lectures on CNNs and Diffusion Models in CSE 573: Computer Vision course at UB.
- Instructor: Prof. Junsong Yuan.
- 🎉 Recipient of CSE Graduate Teaching Award for my Summer CVIP course at UB.
- 🎉 Passed Oral Qualifying Examination at UB!
- 👨🏻🏫 Taught CSE 4/573: Computer Vision and Image Processing at UB, Summer 2025.
-
🎉 Your Text Encoder can be an Object-Level Watermarking Controller is accepted at ICCV 2025!
- (with: Adobe Research, Prof. Siwei Lyu).
-
🎉 Protecting against Unauthorized Latent Diffusion Personalization through Trajectory Shifted Perturbations is accepted at ACM MM 2025!
- (with: Adobe Research, CVG at UMBC).
Contact#
301C Davis Hall, Buffalo NY, 14260