Summary
Soumye Singhal is an Applied Research Scientist with a decade of experience focused on large language models, embeddings, and NLP applications, currently advancing post-training methods at NVIDIA. She combines strong academic roots from IIT Kanpur and Université de Montréal with hands-on research at Mila under advisors including Aaron Courville and Yoshua Bengio. Her work spans student-teacher models, retrieval-augmented domain adaptation for neural machine translation, and cross-modal representation learning, reflecting a rare blend of theoretical depth and applied system building. Based in Old Toronto, she brings both industry-scale model engineering and published research experience to bridge cutting-edge LLM techniques with practical conversational AI products. An understated strength is her continuity across internships and research roles that progressively moved from reinforcement learning and vision-language work into state-of-the-art LLM post-training.
10 years of coding experience
3 years of employment as a software developer
CBSE, Science, CBSE, Science at Bal Bharati Public School, Pitampura
Indian Institute of Technology Kanpur
Master's degree, Computer Science, Master's degree, Computer Science at Université de Montréal
Hindi, English