Abhishek Dalvi

I'm currently working as an AI Research Scientist at LLM Controls Inc., where I am responsible for developing novel RAG systems, along with RL and fine-tuning of LLMs and VLMs for specialized domains.

I completed my Ph.D. in Computer Science and Engineering at Penn State University, University Park, where I was advised by Prof. Vasant Honavar.

At Penn State, I was a member of the Artificial Intelligence Research Laboratory at The College of Information Sciences and Technology. My Ph.D. dissertation, "Hyperdimensional Computing for Graph Representation Learning, Causal Inference, and Image Captioning," explored Hyperdimensional Computing (HDC, also known as Vector Symbolic Architectures) where I introduced novel HDC algorithms as compute-friendly alternatives to deep learning: all of the algorithms use binary representations, require no backpropagation, and learn in a single pass through the data.

Prior to joining Penn State, I completed my M.S. in Computer Science and Engineering at the University at Buffalo, SUNY, advised by Dr. Jing Gao and Dr. Ayan Acharya.

Education

Experience

Research Updates

• Hyperdimensional Cross-Modal Alignment of Frozen Language and Image Models for Efficient Image Captioning:- Preprint available on arXiv.
• C-HDNet: A Fast Hyperdimensional Computing Based Method for Causal Effect Estimation from Networked Observational Data:- Paper accepted at ASONAM 2025 - SNAM (Social Network Analysis and Mining) Journal Track.
• Hyperdimensional Representation Learning for Node Classification and Link Prediction:- Paper accepted at the ACM International Conference on Web Search and Data Mining (WSDM) 2025.
• Causal Effect Estimation Using Random Hyperplane Tessellations:- Paper accepted at the Conference on Causal Learning and Reasoning (CLeaR) 2024.
• Variational Graph Auto-Encoders for Heterogeneous Information Networks:- Paper accepted at a NeurIPS workshop, 2022.
• Best M.S. Thesis Award, CSE Department, University at Buffalo, SUNY. 2021