I am Kshitij Ambilduke, a second-year master’s student at the MVA program, ENS Paris-Saclay. My research interests broadly surround Natural Language Processing (NLP) and I am particularly excited about extensions of NLP towards other modalities like speech and vision. I am currently interning at NAVER LABS Europe under the supervision of Marcely Zanon Boito, Vassilina Nikoulina and Caroline Brun, working on the evaluation of Speech LLMs.
Before my master's, I worked on extending LLMs to speech as a Research Assistant under the guidance of Prof. André F. T. Martins and Prof. Bruno Martins at the SARDINE lab in Lisbon. More recently, I investigated the gains of using complementary multimodal data over unimodal counterparts in a video-game corpus under the supervision of Dr. Djamé Seddah at the ALMAnaCH team, INRIA-Paris.
I completed my bachelor’s in Electronics and Communication from Visvesvaraya National Institute of Technology (VNIT), India. I did my bachelor's thesis with Dr. Anamika Singh where I worked on image captioning. During my pre-final year summer, I worked on interpretable Visual Question Answering under the supervision of Prof. André F. T. Martins and Prof. Bruno Martins.
I was also fortunate to be a part of IvLabs, the AI and Robotics lab of VNIT, where I developed most of my research interests thanks to the oversight of Prof. Shital Chiddarwar
:ligne-icône-de-résumé-cv-103642523--1-: Résumé
:linkdin: KshitijAmbilduke
:gscholar: KshitijAmbilduke
:github: KshitijAmbilduke
$$ \color{337ea9}\rule{630px}{5px} $$
From TOWER to SPIRE: Adding the Speech Modality to a Text-Only LLM
Kshitij Ambilduke, Ben Peters, Sonal Sannigrahi, Anil Keshwani, Tsz Kin Lam, Bruno Martins, Marcely Zanon Boito, André F.T. Martins, Findings of EMNLP 2025
Preprint: https://arxiv.org/abs/2503.10620
Attending to Transforms: A Survey on Transformer-based Image Captioning
Kshitij Ambilduke, Thanmay Jayakumar, Luqman Farooqui, Himanshu Padole, Anamika Singh, Paradigm Shifts in Communication, Embedded Systems, Machine Learning and Signal Processing (PCEMS 2023)