Rohit Salla

Hello! I am a Master's student in Computer Engineering at Virginia Tech, where I work on reinforcement learning for language models and robot control at TEALab with Simon Stepputtis.

Before Virginia Tech, I worked across three labs during my undergrad. At NCCU Taiwan with Prof. Chih-Yi Chiu on token quantization for late-interaction retrieval. At IISc Bangalore with Prof. Suresh Sundaram on object detection transformers for small objects. At CMU with Prof. Min Xu on 3D segmentation and volumetric scene understanding.

Email CV Scholar GitHub LinkedIn

Rohit Salla

Rohit Salla
§ I — Collaborations

Places & People

Virginia Tech
2025 — Present
MS · TEALab
TEALab
w/ S. Stepputtis
Robot Learning
Carnegie Mellon
w/ Prof. Min Xu
3D Segmentation
IISc Bangalore
w/ S. Sundaram
Detection · CV
NCCU Taiwan
w/ C.-Y. Chiu
Quantization
§ II — Updates

News & Events

Jun 2026
Our paper TopoFuse: Topology-Aware Tri-Planar Fusion for 3D Cryo-Electron Tomography Segmentation accepted at ECCV 2026. Work done under the guidance of Prof. Min Xu (CMU) — to be presented in Malmö, Sweden, September 8–11.
Jun 2026
Our paper What Does Disagreement Mean? A Semantic Identifiability Frontier for Pairwise Preference Learning accepted Spotlight at the 2nd Workshop on Epistemic Intelligence in Machine Learning (EIML @ ICML 2026). Oral presentation on July 10.
May 2026
Our paper Learning When to Stop: Prefix-Optimal Dynamic Diffusion Policies for Continuous Control accepted at RLC 2026, to be held in Montréal, Canada, August 14–19. Work done under the guidance of Prof. Simon Stepputtis.
Apr 2026
Our paper CrossQ: Task-Aligned Cross-Token Conditional Quantization for Late Interaction Retrieval accepted at ICML 2026 to be held in Seoul, South Korea in July.
Apr 2026
Our paper AURORA: Neuro-Symbolic Continual Indexing for Evolving RAG Systems accepted at ACL 2026 (Findings) — 64th Annual Meeting of the Association for Computational Linguistics.
Apr 2026
Travelled to ICLR 2026 in Rio de Janeiro, Brazil to present our workshop paper Quantifying Cross-Query Contradictions in Multi-Query LLM Reasoning at the Workshop on Logical Reasoning of Large Language Models. $750 USD Travel Grant · Virginia Tech
Jan 2026
Travelled to AAAI 2026 in Singapore to present our workshop paper Beyond Hallucinations: A Composite Score for Measuring Reliability in Open-Source LLMs at the Workshop on Assessing and Improving Reliability of Foundation Models. $1200 USD Grant · AAAI
Dec 2025
Awarded the TFP Grant of $500 USD from Virginia Tech to attend AAAI 2026.
Oct 2025
Joined TEA Lab to do research in robotics (Diffusion Policy) with Prof. Simon Stepputtis.
Aug 2025
Started my Master's at Virginia Tech.
Mar 2025
Started a research internship at IISc Bangalore under Prof. Suresh Sundaram, working on Detection Transformers for small objects.
Dec 2024
Started a research internship at NCCU Taiwan under Prof. Chih-Yi Chiu, working on multimodal recommendation systems and quantization.
§ III — Selected Publications
— 2026 · Conference Papers
NEURO-SYMBOLIC EVOLVING RAG Q CONTINUAL · t → ∞
Paper 01

AURORA: Neuro-Symbolic Continual Indexing for Evolving RAG Systems

Association for Computational Linguistics — ACL 2026 (Findings) Accepted
Q × D fp32 QUANTIZE 0xA0xC0x70xF int4
Paper 02

CrossQ: Task-Aligned Cross-Token Conditional Quantization for Late Interaction Retrieval

International Conference on Machine Learning — ICML 2026 Accepted
TRI-PLANAR CRYO-ET XY XZ YZ FUSION TOPOLOGY
Paper 03

TopoFuse: Topology-Aware Tri-Planar Fusion for 3D Cryo-Electron Tomography Segmentation

European Conference on Computer Vision — ECCV 2026 Accepted
π(a|s, t) t = T → 0 STOP t*
Paper 04

Learning When to Stop: Prefix-Optimal Dynamic Diffusion Policies for Continuous Control

Reinforcement Learning Conference — RLC 2026 Accepted
V* GREEDY LSVI
Paper 05

When Can Pure Exploitation Succeed in Linear RL? Decoys and Self-Identifiability for Greedy LSVI

Reinforcement Learning Conference — RLC 2026 Accepted
— 2026 · Workshop Papers
PAIRWISE PREF FRONTIER A B ? IDENTIFIABILITY ★
Paper 06

What Does Disagreement Mean? A Semantic Identifiability Frontier for Pairwise Preference Learning

EIML @ ICML 2026 — Workshop on Epistemic Intelligence in Machine Learning Workshop Spotlight
PROMPT Q₁ Q₂ Q₃ YES NO YES CONTRADICT 66%
Paper 07

Quantifying Cross-Query Contradictions in Multi-Query LLM Reasoning

ICLR 2026 Workshop on Logical Reasoning of Large Language Models Workshop Accepted
§ IV — Academic Service
Reviewer
ICLR 2026 CVPR 2026 ICML 2026
§ V — Visitors

Around the World

Always open to talk.  I'd love to chat about AI or to collaborate on projects — feel free to reach out for a virtual coffee chat. The fastest way is sallarohit1011@gmail.com.