🎓About Me
Experienced ML Engineer and Researcher with a Ph.D. in Privacy-preserving AI, 6+ years of research experience in machine learning, deep learning and 5+ years of hands-on experience in developing and deploying machine learning models. Specialized in NLP applications, deep learning, and data science, with a strong background in privacy and security considerations for AI systems. Proven track record in implementing innovative solutions for complex ML problems, particularly in the areas of privacy-preserving machine learning and natural language processing. Motivated team player with analytical thinking, sharp communication skills, and a dedication to promoting secure and ethical data-driven solutions.
- Machine Learning
- Natural Language Processing
- LLM and Generative AI
- Federated Learning
- Deep Learning
- Privacy Preserving Learning
- Artificial Intelligence
- Computational Linguistics
PhD Computer Science
University of Lille
MSc Computer Science
Saarland University
BTech Information Technology
West Bengal University of Technology
My primary research is focused on advancing Natural Language Processing, particularly in speech synthesis systems through neural architectures. At DFKI, I developed:
- An end-to-end neural speech synthesis system integrating Grapheme-to-Phoneme translation and Tacotron-based IPA-to-Speech synthesis
- Multi-speaker, multilingual speech synthesis capabilities for MaryTTS
- Novel unit-selection based speech synthesis modules contributing to Blizzard Challenge 2017-18
During my Ph.D. at INRIA’s MAGNET lab, I extended this expertise to Large Language Models, specifically:
- Implementing decentralized collaborative computation platforms suitable for LLM training
- Developing sophisticated systems for model deployment and data processing
- Creating frameworks for efficient collaborative learning in large-scale language models
My work in privacy preservation encompasses of utilizing Differential Privacy and optimization techniques for privacy-constrained ML systems. Developing transparent privacy measures for decentralized ML while preserving meaningful analytics.
Along with these, I spend my time implementing these protocols across various applications, focusing on the intersection of NLP, LLMs, and privacy-preserving machine learning.