I am MTS Research at Collinear AI. Previously, I worked on diffusion LLMs at Inception and on NLP and multi-modal machine learning at Adobe Research.
I received my PhD from the Indian Institute of Science, Bangalore, advised by Prof. Partha Talukdar, working on knowledge graph question answering (thesis). Before that, I studied Computer Science at BITS Pilani, graduating in 2015.
Research
My research focuses on efficient language models and grounded generation.
I wrote Prompt Lookup Decoding, now integrated into Hugging Face Transformers and vLLM. My work on attribution shipped in Adobe Acrobat’s AI Assistant.
A full publication list is on Google Scholar.
Projects
Everything 3D
Custom 3D-printed figurines and keepsakes made from your photos.
TokenPath.ai
An API that traces AI-generated text back to its sources, built from my research on attribution.
Experiments
Knowledge Cutoff Benchmark
A benchmark measuring the actual knowledge cutoff of LLMs.
Retrieval vs Recall
An experiment testing whether agentic search benchmarks measure retrieval or recall. Published on the Inception blog.
The Seen and the Unseen — Transcripts
Searchable transcripts of 300+ episodes of Amit Varma’s podcast The Seen and the Unseen.
Selected publications
Saxena A. Prompt Lookup Decoding.
Developed a method to speed-up LLM decoding, integrated in transformers and vLLM.
D.J. Bajpai, S. Agarwal, A. Saxena, K. Kulkarni, S. Mitra & M.K. Hanawal. “FlowCast: Trajectory Forecasting for Scalable Zero-Cost Speculative Flow Matching”. International Conference on Learning Representations (ICLR 2026).
S. Somasundaram, A. Phukan & A. Saxena. “PLD+: Accelerating LLM Inference by Leveraging Language Model Artifacts”. Findings of the Association for Computational Linguistics: NAACL 2025.
A. Phukan, S. Somasundaram, A. Saxena, K. Goswami & B.V. Srinivasan. “Peering into the mind of language models: An approach for attribution in contextual question answering”. Findings of the Association for Computational Linguistics: ACL 2024, 11481-11495. Shipped in Adobe Acrobat’s AI Assistant.
Saxena A., Kochsiek A. & Gemulla R. “Sequence-to-Sequence Knowledge Graph Completion and Question Answering”. Accepted to the 2022 Annual Conference of the Association for Computational Linguistics (ACL 2022).
Saxena A., Chakrabarti S. & Talukdar P. “Question Answering Over Temporal Knowledge Graphs”. Accepted to the 2021 Annual Meeting of the Association for Computational Linguistics (ACL 2021).
Saxena A., Tripathi A. & Talukdar P. “Improving Multi-hop Question Answering over Knowledge Graphs using Knowledge Base Embeddings”. Accepted to the 2020 Annual Conference of the Association for Computational Linguistics (ACL 2020).
Work Experience
Collinear AI
MTS Research — November 2026 - Present
Inception, Bangalore
Member of Technical Staff — September 2025 - November 2026
Training and inference for the Mercury family of diffusion LLMs.
Adobe Research, Bangalore
Research Scientist — June 2022 - August 2025
LLMs for document understanding, generation and grounding in the Document Experiences Lab.
Google, Hyderabad
Software Engineer, Tools and Infrastructure — May 2016 - Mar 2017
PayPal, Chennai
Software Engineer 1 — July 2015 - November 2015
Teaching
Guest lectures on LLM decoding at IISc in Prof. Danish Pruthi’s Introduction to NLP (DS 207), 2025 and 2026.
CV
Contact
apoorv@collinear.ai
apoorvumang [at] gmail.com