I am currently working as an Applied Scientist II at Glance, where I develop personalized fashion intelligence using LLMs, VLMs, and user interaction data. My work spans mining user engagement signals to generate synthetic preference data for fine-tuning open-source LLMs, developing semantic ID representations for LLM-based generative recommendation, and migrating production pipelines from proprietary APIs to self-hosted open-source LLMs.
Previously, I spent 3.5 years as a Research Associate II at the Multimodal Content Experiences Lab at Adobe Research, where I primarily worked with Dr. Apoorv Saxena and Dr. Balaji Srinivasan on leveraging Large Language Models (LLMs)/Multimodal Large Language Models (MLLMs) for document experience projects for Adobe Acrobat and Adobe Express. I worked across a wide range of research areas — retrieval and attribution for document question answering, document stylization and transformation, graphic design generation, and speculative decoding — publishing at top NLP conferences (EMNLP, ACL, EACL, NAACL) and filing multiple patents. Notably, I co-developed a training-free, token-level attribution algorithm with colleagues that leverages the contextual nature of LLM embeddings (published at ACL Findings 2024), and co-developed its integration for contract entity attribution in the Adobe Acrobat AI Assistant — a capability highlighted by Adobe’s CEO in a Fox Business interview.
A common thread across this work has been using LLM internals — hidden states, attention, and activations — to understand and steer model behavior, most recently in ContextFocus, an activation steering method for improving context faithfulness in LLMs. I’m increasingly drawn to this direction and am looking to deepen my work in AI safety and interpretability research.
I completed my bachelor’s thesis under the supervision of Prof. N Venkateswaran at SSN College of Engineering. My project focused on road object detection from radar sensor data using machine learning and deep learning object detection techniques. During my undergraduate studies, I also explored the estimation of tracer kinetic parameters from undersampled DCE-MRI data, under the supervision of Dr. Phaneendra Yalavarthy at the Medical Imaging Lab, Indian Institute of Science, Bangalore.
If you’d like to know more about my work or discuss potential collaborations, please check out my CV. I’m always open to new opportunities and interesting conversations!
| Mar 24, 2026 | Our patent on evidence retrieval for long-document QA was officially granted: US Patent 12,585,685! |
| Jan 22, 2026 | Joined Glance as an Applied Scientist II, working on personalized fashion intelligence. |
| Jan 07, 2026 | Posted a new preprint, ContextFocus — an activation steering method for keeping LLMs faithful to their given context. Check it out on arXiv. |
| Apr 29, 2025 | PLD+, our paper on speeding up LLM inference using model artifacts, was published at NAACL 2025 (Findings) in Albuquerque. |
| Feb 25, 2025 | PostDoc, our work on turning long documents into posters using deep submodular optimization, was published at AAAI 2025 in Philadelphia. |
| Feb 15, 2025 | The contract entity attribution technology I helped build for Adobe Acrobat AI Assistant was highlighted by Adobe’s CEO in a Fox Business interview. |
| Aug 11, 2024 | Our attribution paper — using LLM hidden states to trace generated answers back to their source text — was accepted to ACL 2024 (Findings) in Bangkok. |
| Mar 17, 2024 | Co-authored a paper on generating persona-aware slides from documents with LLMs, published at EACL 2024 in St. Julian’s, Malta. |
| Dec 06, 2023 | Presented our paper on discourse-guided evidence retrieval for long documents at EMNLP 2023 (Findings) in Singapore. |
Publications
2026
-
ContextFocus: Activation Steering for Contextual Faithfulness in Large Language Models
Nikhil Anand, Shwetha Somasundaram, Anirudh Phukan, and 2 more authors
arXiv preprint arXiv:2601.04131, 2026
Large Language Models (LLMs) encode vast amounts of parametric knowledge during pre-training. As world knowledge evolves, effective deployment increasingly depends on their ability to faithfully follow externally retrieved context. When such evidence conflicts with the model’s internal knowledge, LLMs often default to memorized facts, producing unfaithful outputs. In this work, we introduce ContextFocus, a lightweight activation steering approach that improves context faithfulness in such knowledge-conflict settings while preserving fluency and efficiency. Unlike prior approaches, our solution requires no model finetuning and incurs minimal inference-time overhead, making it highly efficient. We evaluate ContextFocus on the ConFiQA benchmark, comparing it against strong baselines including ContextDPO, COIECD, and prompting-based methods. Furthermore, we show that our method is complementary to prompting strategies and remains effective on larger models. Extensive experiments show that ContextFocus significantly improves contextual-faithfulness. Our results highlight the effectiveness, robustness, and efficiency of ContextFocus in improving contextual-faithfulness of LLM outputs.
2025
-
Deep Submodular Optimization and LLM for Multimodal Content Extraction and Automatic Poster Generation from Long Document
Vijay Jaisankar, Sambaran Bandyopadhyay, Kalp Vyas, and 2 more authors
In Proceedings of the AAAI Conference on Artificial Intelligence, 2025
A poster from a long input document can be considered as a one-page easy-to-read multimodal (text and images) summary presented on a nice template with good design elements. Automatic transformation of a long document into a poster is a very less studied but challenging task. It involves content summarization of the input document followed by template generation and harmonization. In this work, we propose a novel deep submodular function which can be trained on ground truth summaries to extract multimodal content from the document and explicitly ensures good coverage, diversity and alignment of text and images. Then, we use an LLM based paraphraser and propose to generate a template with various design aspects conditioned on the input content. We show the merits of our approach through extensive automated and human evaluations.
2024
-
PLD+: Accelerating LLM inference by leveraging Language Model Artifacts
Shwetha Somasundaram, Anirudh Phukan, and Apoorv Saxena
arXiv preprint arXiv:2412.01447, 2024
To reduce the latency associated with autoretrogressive LLM inference, speculative decoding has emerged as a novel decoding paradigm, where future tokens are drafted and verified in parallel. However, the practical deployment of speculative decoding is hindered by its requirements for additional computational resources and fine-tuning, which limits its out-of-the-box usability. To address these challenges, we present PLD+, a suite of novel algorithms developed to accelerate the inference process of LLMs, particularly for input-guided tasks. These tasks, which include code editing, text editing, summarization, etc., often feature outputs with substantial overlap with their inputs-an attribute PLD+ is designed to exploit. PLD+ also leverages the artifacts (attention and hidden states) generated during inference to accelerate inference speed. We test our approach on five input-guided tasks and through extensive experiments we find that PLD+ outperforms all tuning-free approaches. In the greedy setting, it even outperforms the state-of-the-art tuning-dependent approach EAGLE on four of the tasks. (by a margin of upto 2.31 in terms of avg. speedup). Our approach is tuning free, does not require any additional compute and can easily be used for accelerating inference of any LLM.
-
Peering into the Mind of Language Models: An Approach for Attribution in Contextual Question Answering
Anirudh Phukan, Shwetha Somasundaram, Apoorv Saxena, and 2 more authors
In Findings of the Association for Computational Linguistics ACL 2024, 2024
With the enhancement in the field of generative artificial intelligence (AI), contextual question answering has become extremely relevant. Attributing model generations to the input source document is essential to ensure trustworthiness and reliability. We observe that when large language models (LLMs) are used for contextual question answering, the output answer often consists of text copied verbatim from the input prompt which is linked together with “glue text” generated by the LLM. Motivated by this, we propose that LLMs have an inherent awareness from where the text was copied, likely captured in the hidden states of the LLM. We introduce a novel method for attribution in contextual question answering, leveraging the hidden state representations of LLMs. Our approach bypasses the need for extensive model retraining and retrieval model overhead, offering granular attributions and preserving the quality of generated answers. Our experimental results demonstrate that our method performs on par or better than GPT-4 at identifying verbatim copied segments in LLM generations and in attributing these segments to their source. Importantly, our method shows robust performance across various LLM architectures, highlighting its broad applicability. Additionally, we present Verifiability-granular, an attribution dataset which has token level annotations for LLM generations in the contextual question answering setup.
-
Presentations by the Humans and For the Humans: Harnessing LLMs for Generating Persona-Aware Slides from Documents
Ishani Mondal, Shwetha S, Anandhavelu Natarajan, and 3 more authors
In Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), Mar 2024
Scientific papers and slides are two different representations of the same underlying information, but both require substantial work to prepare. While there had been prior efforts on automating document-to-slides generation, there is still a pressing need of customizing the presentation of content aligning with the persona of target audience or duration of presentation. This paper first introduces the concept of end-user specification-aware document to slides conversion that incorporates end-user specifications into the conversion process. For this, we initially introduce a new dataset reuse the existing SciDuet dataset consisting of pairs of papers and corresponding slides decks from recent years’ *ACL conferences to create four persona-aware configurations. Secondly, we present Persona-Aware-D2S, a novel approach by finetuning LLMs using target audience feedback to create persona-aware slides from scientific documents. Our evaluation on both automated metrics and qualitative human evaluation suggests that by incorporating end-user specifications into the conversion process, our model can create presentations that are not only informative but also tailored to expectations and cognitive abilities of target audience.
2023
-
Drilling Down into the Discourse Structure with LLMs for Long Document Question Answering
Inderjeet Nair*, Shwetha Somasundaram*, Apoorv Saxena, and 1 more author
In Findings of the Association for Computational Linguistics: EMNLP 2023, Mar 2023
We address the task of evidence retrieval for long document question answering, which involves locating relevant paragraphs within a document to answer a question. We aim to assess the applicability of large language models (LLMs) in the task of zero-shot long document evidence retrieval, owing to their unprecedented performance across various NLP tasks. However, currently the LLMs can consume limited context lengths as input, thus providing document chunks as inputs might overlook the global context while missing out on capturing the inter-segment dependencies. Moreover, directly feeding the large input sets can incur significant computational costs, particularly when processing the entire document (and potentially incurring monetary expenses with enterprise APIs like OpenAI’s GPT variants). To address these challenges, we propose a suite of techniques that exploit the discourse structure commonly found in documents. By utilizing this structure, we create a condensed representation of the document, enabling a more comprehensive understanding and analysis of relationships between different parts. We retain 99.6% of the best zero-shot approach’s performance, while processing only 26% of the total tokens used by the best approach in the information seeking evidence retrieval setup. We also show how our approach can be combined with \textitself-ask reasoning agent to achieve best zero-shot performance in complex multi-hop question answering, just ≈4% short of zero-shot performance using gold evidence.
Patents
- Generating Draft Sequence Rankings for Speculative Decoding Using Large Language Model Hidden States (US Patent App. 18/924,398)
- Evidence Retrieval for Long Document Question Answering Using Large Language Models (US Patent 12,585,685, granted)
- Automatic generation of handouts from multi-modal documents (US Patent App. 18/542,161)
- Merging misidentified text structures in a document (US Patent App. 18/511,111)
- Generating targeted layouts from source documents utilizing large language models with semantic hierarchical transformations (US Patent App. 18/809,147)
- Generating a digital poster including multimodal content extracted from a source document (US Patent App. 18/619,667)
- Document-based presentation generation (US Patent App. 18/675,451)
- Context-focused steering for machine learning models (US Patent App. 19/229,734)
