Janvi Palan

I've specialized in the fields of Machine learning and distributed systems, with a focus on natural language processing. Currently working on problems in the generative AI space, especially multi-modal understanding and generation using LLMs - I enjoy writing code and at scale. At Samsung Research - I've been involved with building digital humans that are hyper-realistic and can be used in various marketing, sales and service scenarios! I’m building out the NLP/conversational AI engines to make our experiences with digital humans friendlier and more intelligent. I work with emotional context extraction and enhanced NLU to guide better dialogue between humans and NEONs. I was previously responsible for the generation of realistic facial behavior and expressions and body motion given multi-modal control inputs using various generative AI techniques and motion models. I'm also leading the development and research for a Personal AI Search project, helping users understand, query, retrieve and communicate with their personal on-device data in a privacy-preserving manner. I've done research in the areas of Information retrieval for recommendations and personalization. Some of my projects include identifying echo chambers and semi-supervised online hate speech in social media, using graph neural networks, as well as personalized recommendation systems in a traditional as well as in a reinforcement learning setting. My expertise lies as an end-to-end machine learning researcher/engineer, experimenting with data, modeling and integrating ML pipelines in the entire software development life cycle stack, along with optimization and deployment of these models, coupled with my research exposure in multi-modal machine learning and generative AI. I love contributing to research talks, learning about start-ups and teams making a socio-technical impact, and participating in open source projects, in the fields of AI/ML. Always happy to connect to learn about new, impactful opportunities!