Shambhavi Mishra

Hi! I’m Shambhavi Mishra, a PhD student at École de technologie supérieure (ÉTS), where I work under the guidance of Prof. Jose Dolz and Prof. Marco Pedersoli. My research focuses on advancing Foundation Models for label-scarce learning, with a particular emphasis on improving model reliability, calibration, and adaptation to make AI more robust in real-world scenarios.

As a Visiting Researcher at ServiceNow Research - Foundation Models Lab, I work on enhancing Mixture of Experts (MoEs), SSM/Transformer hybrids, and designing novel model architectures to push the boundaries of large-scale AI systems.

My industry experience spans multiple domains, including patent and legal data analytics, fine-tuning LLMs, and developing AI-driven solutions for complex, domain-specific challenges. At companies like FrandAvenue and NeuroSensum, I have worked on leveraging LLMs for structured document understanding, multilingual text analytics, and automation in intellectual property research. These experiences have strengthened my expertise in adapting foundation models for real-world applications, bridging the gap between research innovation and industry adoption.

I am also passionate about community-driven research discussions. I founded Vision Language Talks, a platform that connects researchers and authors to discuss cutting-edge advancements in AI, fostering collaboration and knowledge-sharing. Additionally, I have organized machine learning workshops for underprivileged communities, combining my passion for education with my commitment to making AI more accessible.

I’m always open to collaborations, discussions, and new challenges. Feel free to explore my work and reach out—let’s connect!