Hi! I am Divyansh, a Senior Undergraduate in Computer Science & Engineering department at IIT Kanpur.
I am interested in learning about new advancements in technologies in general. I am passionate about AI and keep track of the latest advancements via twitter(X now), online articles , papers, and sometimes youtube videos.

I interned at Adobe Research, Bangalore in summer, 2024. I worked on Vision Language Models, curating a multi-modal attribution system for answers generated by AI-assistants. We have US patent accepted for this work and also a paper accepted in NAACL, 2025 main track.

Earlier, In the summer of 2023, I did a research internship at UIUC at Blender Labs where I worked on short video understanding with LLMs as judge.

Updates

The Era of Scaling Test-Time Compute: Reasoning Large Language Models

Blog: Exploring how thought-search during inference improves reasoning capabilities of LLMs.
Divyansh

Coming Soon

Beyond Logit Lens: Contextual Embeddings for Robust Hallucination Detection & Grounding in VLMs, NAACL'25

Paper: Anirudh Phukan, Divyansh , Harshit Kumar Morj, Vaishnavi, Apoorv Saxena, Koustava Goswami

Adobe Research | Summer, 2024

FIQA-Challenges: Aspect based sentiment analysis

Presentation: Fine-tune BERT + Multi-head attention on aspect for sentiment extraction.

Barclays Case Study Competition | Nov'24