PeopleofSuperintelligence

Researcher

Pulkit Tandon

Pulkit Tandon is a researcher whose work covers data compression and data selection for machine learning. After a bachelor's degree in technology at IIT Bombay, he completed a Stanford PhD in electrical engineering, interned at Netflix on video compression, and coauthored the 2021 Txt2Vid pipeline, which compresses talking-head video to a text transcript. In 2023, he coauthored a paper with Germain Kolossov and Andrea Montanari on data selection under weak supervision, which was an oral at ICLR 2024 and won an Honorable Mention. He is a researcher at Granica.ai, and in 2024 he also served as an adjunct lecturer at Stanford, teaching data compression.

Now at
Granica.ai

Timeline

  1. By 2021

    Doctoral work at Stanford

    He earns a bachelor's degree in technology at the Indian Institute of Technology, Bombay, and joins Stanford's Electrical Engineering department as a Ph.D. student and a primary student in Tsachy Weissman's group. By 2021, his research covers data compression, neural engineering, perceptual engineering and brain-machine interfaces. He also interns at Netflix, working on deep-learning video compression and objective video quality metrics.

  2. 2021

    Txt2Vid talking-head compression

    In June 2021, the Txt2Vid paper appears on arXiv with Tandon as first author. The pipeline compresses webcam video to a text transcript and rebuilds it with voice-cloning and lip-syncing models, cutting the bitrate by two to three orders of magnitude. He presents it at VideoTech 2021, and he is among the innovators behind the technology, which Stanford files as docket S21-205 and which has US patent 12,567,398.

  3. 2023

    A theory of data selection

    In September 2023, he coauthors "Towards a statistical theory of data selection under weak supervision" with Germain Kolossov and Andrea Montanari. The paper studies picking a smaller subsample of unlabeled data with the help of a surrogate model, labeling it, and training with regularized empirical risk minimization. It finds that data selection can beat training on the full sample in some cases.

  4. 2024

    ICLR oral and honorable mention

    The data-selection paper is presented as an oral at ICLR 2024. It receives an Honorable Mention among the conference awards.

  5. October 2024

    Research engineer at Granica

    In October 2024, he presents the data-selection work at an appliedAI Institute TransferLab seminar as a Research Engineer at Granica, where he works on data selection, compression and optimization for AI applications. He is also an Adjunct Lecturer at Stanford, teaching courses on data compression, and holds a Ph.D. in Electrical Engineering from Stanford.

  6. As of October 2026

    Researcher at Granica.ai

    He is a researcher at Granica.ai with a Stanford PhD. His research interests span data selection, machine learning, compression and neuroengineering.

Career

  1. NowGranica.aiResearcher
  2. From 2024GranicaResearch Engineer
  3. 2024Stanford UniversityAdjunct Lecturer
  4. NetflixIntern

Education

  1. IIT BombayBachelor's degree in technology
  2. Stanford UniversityPh.D. in Electrical Engineering

From the biography.

  1. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Stanford University. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  2. Google had the most people shaping AI every year from 2014 to 2025Includes Stanford University. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  3. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Stanford University. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.
  4. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Stanford University. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Pulkit Tandon yet. Send us a free-licensed photo.

  1. Pulkit Tandon at VideoTech 2021 vtconf.com. Checked 22 October 2026.
  2. Group web.stanford.edu. Checked 22 October 2026.
  3. Txt2Vid: Ultra-Low Bitrate Compression of Talking-Head Videos via Text arxiv.org. Checked 22 October 2026.
  4. Txt2Vid: Compressing Talking-head Videos to Text | Explore Technologies techfinder.stanford.edu. Checked 22 October 2026.
  5. Towards a statistical theory of data selection under weak supervision arxiv.org. Checked 22 October 2026.
  6. ICLR 2024 Awards iclr.cc. Checked 22 October 2026.
  7. Towards a statistical theory of data selection under weak supervision | TransferLab — appliedAI Institute transferlab.ai. Checked 22 October 2026.
  8. Pulkit Tandon - Google Scholar scholar.google.com. Checked 22 October 2026.