PeopleofSuperintelligence

Researcher

Hirakendu Das

Hirakendu Das is a researcher in information theory and machine learning whose work covers distribution estimation, pattern maximum likelihood and scalable decision trees. At UC San Diego's ECE department, he won the 2010 Jack Keil Wolf ISIT Student Paper Award for a paper on reconstructing strings from their substring compositions. With Alon Orlitsky, Jayadev Acharya and coauthors, he published papers on closeness testing, classification, profile redundancy and string reconstruction, and by 2014 he worked at Yahoo Labs on the initial decision tree code for Spark MLlib and on click modeling for sponsored search. A 2017 ICML paper with Acharya, Orlitsky and Ananda Theertha Suresh showed that profile maximum likelihood is sample competitive for all symmetric properties of discrete distributions, earning a Best Paper honorable mention.

Timeline

  1. 2010

    ISIT student paper award

    At UC San Diego's ECE department, he wins the 2010 Jack Keil Wolf ISIT Student Paper Award for the paper "On reconstructing a string from its substring compositions".

  2. 2011

    Competitive closeness testing at COLT

    In 2011, he coauthors "Competitive Closeness Testing" at the 24th Annual Conference on Learning Theory with Jayadev Acharya, Ashkan Jafarpour, Alon Orlitsky and Shengjun Pan. The paper gives a pattern-maximum-likelihood algorithm for testing whether two sequences come from the same distribution without assumptions on support size.

  3. 2012

    Classification and profile redundancy bounds

    At the 25th Annual Conference on Learning Theory in 2012, he coauthors "Competitive Classification and Closeness Testing", constructing a classifier and closeness test that use at most O(n^(3/2)) samples to match the best symmetric method. In 2012, he also coauthors the NIPS paper "Tight Bounds on Profile Redundancy and Distinguishability" with Jayadev Acharya and Alon Orlitsky, which places the redundancy of profiles of length-n i.i.d. sequences between 0.3·n^(1/3) and n^(1/3) log2 n.

  4. By 2014

    Decision trees in Spark MLlib

    He works at Yahoo Labs on the initial Spark MLlib decision tree implementation, developed jointly with Evan Sparks of UC Berkeley's AMPLab and Ameet Talwalkar and Xiangrui Meng of Databricks. By September 2014, the distributed decision tree code is part of MLlib from release 1.0, and Spark 1.1 adds multiclass classification and Python APIs.

  5. December 2014

    Sponsored search click modeling

    As a Yahoo Labs researcher, he speaks at UC San Diego's CSE 259 AI seminar in December 2014 on large-scale click modeling for sponsored search.

  6. 2014–2015

    Reconstructing strings from compositions

    In March 2014, he coauthors "String Reconstruction from Substring Compositions" on arXiv with Jayadev Acharya, Olgica Milenkovic, Alon Orlitsky and Shengjun Pan. Motivated by mass-spectrometry protein sequencing, it shows that all strings of length 7, one less than a prime, or one less than twice a prime can be reconstructed uniquely up to reversal, and the paper appears in the SIAM Journal on Discrete Mathematics in 2015.

  7. November 2016

    Profile maximum likelihood preprint

    With Jayadev Acharya, Alon Orlitsky and Ananda Theertha Suresh, he posts the arXiv preprint "A Unified Maximum Likelihood Approach for Optimal Distribution Property Estimation" in November 2016. It proves that one plug-in estimator, profile maximum likelihood, performs as well as the best specialized techniques for support size, support coverage, entropy and proximity to uniformity.

  8. 2017

    ICML honorable mention

    While at Yahoo! in 2017, he coauthors the ICML paper "A Unified Maximum Likelihood Approach for Estimating Symmetric Properties of Discrete Distributions" with Jayadev Acharya, Alon Orlitsky and Ananda Theertha Suresh, earning a Best Paper honorable mention. The paper shows that profile maximum likelihood is sample competitive for all symmetric properties and asymptotically sample-optimal for support size, support coverage, entropy and proximity to uniformity.

  1. Universities keep AI researchers for a median of 6 years, compared with 2 years at frontier labsIncludes Yahoo. Across 15,811 dated work stays (6,529 people), median role tenure runs from 6 years at universities to 2 years at frontier labs. We compare all five organization types, starting decades from the 1970s to the 2020s, and the organizations that keep people longest.
  2. 23% of people shaping AI who left a university for a company later returned to a universityIncludes Yahoo. We estimate that 468 of the 2,011 people shaping AI who left a university for a company (23%) later returned to a university after a median of 5 years, and 163 of 829 (20%) made the reverse round trip.
  3. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Yahoo. 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.
  4. Google had the most people shaping AI every year from 2014 to 2025Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.

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 Hirakendu Das yet. Send us a free-licensed photo.

  1. Member profile #8895 | IEEE Information Theory Society itsoc.org. Checked 22 October 2026.
  2. Competitive Closeness Testing proceedings.mlr.press. Checked 22 October 2026.
  3. Competitive Classification and Closeness Testing proceedings.mlr.press. Checked 22 October 2026.
  4. Tight Bounds on Profile Redundancy and Distinguishability papers.neurips.cc. Checked 22 October 2026.
  5. Scalable Decision Trees in MLlib | Databricks Blog databricks.com. Checked 22 October 2026.
  6. CSE 259. UCSD AI Seminar cseweb.ucsd.edu. Checked 22 October 2026.
  7. [1403.2439] String Reconstruction from Substring Compositions arxiv.org. Checked 22 October 2026.
  8. Hirakendu Das - Google Scholar scholar.google.com. Checked 22 October 2026.
  9. [1611.02960] A Unified Maximum Likelihood Approach for Optimal Distribution Property Estimation arxiv.org. Checked 22 October 2026.
  10. ICML 2017 Awards icml.cc. Checked 22 October 2026.
  11. ICML 2017 Accepted Papers icml.cc. Checked 22 October 2026.
  12. A Unified Maximum Likelihood Approach for Estimating Symmetric Properties of Discrete Distributions proceedings.mlr.press. Checked 22 October 2026.