| CARVIEW |
I am an ivado postdoctoral researcher at MILA, working with Sarath Chandar. My recent research interests span vision-language-action (VLA) world models, (continual/temporal) learning on streaming data, and real-world verification of generative model outputs.
I completed my PhD at UNSW Sydney in August 2025, where I was advised by Lina Yao and Dong Gong. During the latter half of my PhD, I worked as an applied research scientist at openstream.ai, and as a research intern at Sony (hosted by Shiqi Yang and Shusuke Takahashi ) and Tencent (hosted by Shengju Qian).
Prior to my PhD, I worked on continual learning with Joost van de Weijer, and did my Erasmus Mundus Joint Master's Degree (EMJMD) in Advanced Systems Dependability at the University of St Andrews, the UK and l'Université de Lorraine, France. During my master's, I interned in Emmanuel Vincent's group at Inria Nancy. I once wrote this medium blog documenting my EMJMD experience to help guide future aspirants.
In my free time, I enjoy snorkelling, hiking, and binge-watching. I grew up in eastern Nepal, and every couple of years, I like to plan week-long treks in the Nepalese Himalayas (hit me up if you want to join in the next one!).
news
- Dec 2025: I will be serving in the steering committee of a **new** mila talk series led by Hugo Larochelle.
- Oct 2025: Quoted in this MIT news story for perspectives on personalized object localization.
- Aug 2025: My PhD thesis was recommended for the Dean's Award for outstanding PhD theses at UNSW Sydney.
- Jun 2025: Presented my paper "Mining Your Own Secrets" at the Sydney AI meetup .
- May 2025: Mentoring session + Talk on continual learning in Bishesh Khanal's group, NAAMII, Nepal.
- Apr 2025: Gave a virtual talk on continual learning in Marinka Zitnik's group, Harvard University.
- Apr 2025: Attending ICLR'25 in Singapore to present “Mining your Own Secrets” paper.
- Dec 2024: Presented my paper "CLAP4CLIP" at NeurIPS 2024 in Vancouver.
- Nov 2024: Won the Tiktok-sponsored Best Student Presentation award for "CLAP4CLIP" at 2024 Sydney AI meetup.
- Jul 2024: Presented my paper "CLAP4CLIP" at the 2nd Bayes duality workshop, RIKEN AIP, Tokyo.
- Apr 2024: Won a travel grant for presenting my paper "NPCL" at the EEML summer school, Novi Sad, Serbia.
Experience
IVADO postdoctoral fellow (Sep 2025 - Present)
MILA • Montréal, Canada 🇨🇦
Research aligned with advancing Canada's R3AI initiative.
Applied AI Scientist (May 2025 - Aug 2025)
OpenStream.ai • Melbourne, Australia 🇦🇺
Infra-focus: Developed production-grade conversational LLM agents for enterprise clients.
ML-focus: Implemented & shipped a POC for neuro-symbolic verification of multi-agent systems.
AI Research Intern (Sep 2024 - Mar 2025)
LightSpeed Studios, Tencent • Sydney, Australia 🇦🇺
Worked on controllable image generation and preference optimization for multi-modal LLMs.
Research Scientist Intern (May 2024 - Aug 2024)
Creative AI Lab, Sony Group Corporation • Tokyo, Japan 🇯🇵
Worked on continual personalization of pre-trained text-to-image diffusion models.
Research Assistant (Sep 2021 - Jan 2022)
Computer Vision Centre, Universitat Autònoma de Barcelona • Barcelona, Spain 🇪🇸
Worked on rehearsal-free continual learning for Vision Transformers (ViTs).
Research Intern (Mar 2021 - Jul 2021)
Multispeech group, Inria Nancy • Nancy, France 🇫🇷
Worked on learning domain-specific language models for speech recognition.
Machine Learning Engineer (Jun 2018 - Jul 2019)
FactSet Research Systems Inc. • Hyderabad, India 🇮🇳
Worked on improving FactSet's named entity recognition service with acronym disambiguation and neural topic modeling.
Awards & Recognition
- IVADO Postdoctoral Fellowship: Among the 11 recipients of the 2025 cohort.
- CSE writing fellowship: From UNSW Sydney, in recognition of a strong PhD publication record (2025).
- Tiktok-sponsored Best Student Presentation awardee at the Sydney AI meetup (2024).
- Travel grant awardee for Eastern European Machine Learning (EEML) summer school, Serbia (2024).
- Best runner-up paper awardee at the CVPR 2022 workshop on Continual Learning.
- University International Postgraduate Award (UIPA) recipient for PhD studies at UNSW Sydney (2022).
- Best master’s thesis award for Erasmus+ DEPEND 2019-21 cohort.
- Best students’ poster at Digital Ethics4EU 2021 workshop, TU Dublin.
- Winner of Barclays chatbot challenge at Hack the Burgh 2020, the University of Edinburgh.
- Erasmus Mundus scholarship for joint Master’s degree studies in the UK and France (2019).
Academic Services
- Reviewer for:
- World Modeling Workshop (WMW 2026)
- International Conference on Learning Representations (ICLR 2024, 2025, 2026)
- IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR 2024, 2025, 2026)
- Conference on Neural Information Processing Systems (NeurIPS 2023, 2024)
- IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2026)
- IEEE Transactions on Image Processing (TIP)
- IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)
- Program Committee (PC) member for:
- Industry Track of the Web Conference (WWW 2025, 2026)
- European Conference on Artificial Intelligence (ECAI 2025)
- Workshop Proposals, Conference on Information & Knowledge Management (CIKM 2023)
tutoring at UNSW
- COMP6713 (Natural Language Processing), taught by Aditya Joshi [also wrote section 6 "Coding Assessments" of the ACL workshop paper on our teaching methodology & findings]
- COMP9418 (Advanced Topics in Statistical Machine Learning), taught by Gustavo Batista
- ZZEN9444 (Neural Networks and Deep Learning), taught by Dong Gong
Selected Publications
Probing the Effectiveness of World Models for Spatial Reasoning through Test-Time Scaling
Saurav Jha, M. Jehanzeb Mirza, Wei Lin, Shiqi Yang, Sarath Chandar
We propose Verification through Spatial Assertions (ViSA), a proposer-solver method that enables faithful test-time verification of world model views for enhancing the spatial reasoning in existing VLMs.
Mining Your Own Secrets: Diffusion Classifier Scores for Continual Personalization of Text-to-Image Diffusion Models
Saurav Jha, Shiqi Yang, Masaki Ishii, Meng Zhao, Christian Simon, Jehanzeb Mirza, Dong Gong, Lina Yao, Shusuke Takahashi, Yuki Mitsufuji
ICLR 2025
We propose using diffusion classifier scores for regularizing the parameter-space and function-space of text-to-image diffusion models, to achieve continual personalization.
CLAP4CLIP: Continual LeArning with Probabilistic finetuning for Vision-Language Models
Saurav Jha, Dong Gong, Lina Yao
Our work proposes Continual LeArning with Probabilistic finetuning (CLAP) - a probabilistic modeling frame- work over visual-guided text features per task, thus providing more calibrated CL finetuning.
NPCL: Neural Processes for Uncertainty-Aware Continual Learning
Saurav Jha, Dong Gong, He Zhao, Lina Yao
We propose a neural process-based continual learning approach with task-specific modules arranged in a hierarchical latent variable model. We tailor regularizers on the learned latent distributions to alleviate forgetting.
Towards Exemplar-Free Continual Learning in Vision Transformers: an Account of Attention, Functional and Weight Regularization
Francesco Pelosin*, Saurav Jha*, Andrea Torsello, Bogdan Raducanu, Joost van de Weijer
We investigate the continual learning of Vision Transformers (ViT) for the challenging exemplar-free scenario, with special focus on how to efficiently distill the knowledge of its crucial self-attention mechanism.