Swaroop Mishra

I have been involved in the full development lifecycle of language models for several years, encompassing pre-training, post-training, data quality, human data collection, instruction following, reasoning capabilities, model architecture design, and agent development. Over the last 15 years, my research experience has spanned more than 10 industries and research institutions across various roles, including research scientist, software engineer, and technical consultant.

I am a Research Scientist at Google DeepMind (Ex-Google Brain), where I work on Gemini reasoning. I built a System that got some crucial points in the International Mathematical Olympiad (IMO) 2024 and critically contributed to our Silver Medal along with AlphaProof and AlphaGeometry. I also co-led reasoning research that contributed to various successful launches SOTA Gemini model including the popular Google IO 2024 launch. Most of my patented works at Google revolves around LLM Reasoning, Scaling, Post-Training, Self-Improvement and Inference.

My main research contributions (in the form of publicly available papers) are

• Instruction-tuning: Pioneering work: the first paper on instruction-tuning, Scaling: Super-NaturalInstructions, Applications: Biomedical Instruction-tuning, Aspect Based Sentiment Analysis, Evaluation: Instruction-following Evals, instruction-example equivalence and Bias: instruction-bias (EACL 2023 Outstanding Paper Award)
• Reasoning: Methods: reframing, question-decomposition, math via programs, help me think, learn to explain, take a step back, selective answering, summary of thought and Dataset: NumGLUE, FeasibilityQA and TarGEN
• Model Self-Improvement: Training Method: Self-Instruct, Evaluation Method: Self-Discover and Limitation: Self-Correction Limitations.
• New Products and Beyond: User-friendly and controllable LLMs for structured thoughts and personalized model responses and other patented works.

I received the GDM Impact Award 2025 for my contribution to the silver medal achieving model in IMO 2024. I am also honored to be awarded the Odia Bhasa Samman 2025 (Odia Language Award 2025). I also got the ASU Dean’s Dissertation Award 2023 (1 in the school) and CEN-CS Outstanding Student Award 2023 (1 in the department) for my PhD Dissertation: "Towards Development of Models that Learn New Tasks from Instructions". The award video summarizes the contribution of my dissertation to ChatGPT and other LLMs. My work "Natural Instructions" received the AI2 "Lasting Impact Paper Award" after 3.5 years of the first publication date. I am a co-organizer of the MATH-AI workshop at NeurIPS 2022 and NeurIPS 2024.

[Selected Media Coverage]:

• The New York Times article covering our IMO-Silver (group photo in the article)
• A popular local media Interview
• Kanak News Interview
• A post on The Making of a Scientist: From Classroom Rebel to AI Innovator
• A podcast on How to get started on a Research Paper? Interview with Swaroop Mishra: Part 1 & Part 2
• A long technical podcast on Instruction Tuning, Prompt Engineering, & Self Improving LLMs
• A short podcast in NIST Podcast Series on Powering ChatGPT, Gemini & LLMs: Insights from Dr. Swaroop Mishra, Google.
• A YouTube video on my journey
• Ones Changing The World - 1CW Podcast
• Swaroop interviewing Jeff Dean at the Odisha AI Conference 2024: Video
• Swaroop mentioned by Vishal Gondal on the Neem Karoli Baba Podcast. Here is the full Podcast Video

[Selected Panel Discussions and Invited Talks]:

• NeurIPS 2024 Panel Discussion with Yoshua Bengio and group hosted by Zhijing Jin: Here is a Photo
• MATH-AI workshop Panel at NeurIPS 2024 with Noam Brown, Dawn Song, Jeremy Avigad and Junehyuk Jung : Here is a Photo
• ACL 2023 Panel Discussion with Sara Hooker, Danqi Chen
. Here is a Photo
• Invited Talk at Allen AI [2021]
• Invited Talk at IIT Kanpur [2021]
• Invited Talk at University of Toronto [2022]
• Invited Talk at Google Research [2022]
• Invited Talk at Facebook AI Research (FAIR) [2022]
• Invited Talk at Microsoft Research [2022]
• Invited Talk at Stanford [2022]
• Invited Talk at Intel [2023]
• Invited Talk at IIT Kharagpur [2023]
• Invited Talk at Oracle Research [2023]
• Invited Talk at Silicon Bhubaneswar [2023] Photo
• Invited Talk at MIT [2024]: Photo 1, 2
• Invited Talk at IIT Mandi [2024]: Photo
• Invited Talk at IIT Bhubaneswar [2024]
• Invited Talk at KIIT Bhubaneswar [2024]
• Invited Talk at Xavier Univeristy Bhubaneswar [2024]
• Invited Talk at NIST Berhampur [2024] Photo
• AAAI 2025 Panel Discussion
• Invited Talk at NIT Silchar [2025] at the Neurathan

[Research Experience]:

Follow @Swarooprm7  /  Email  /  Google Scholar /  Linkedin /  Instagram /  Facebook