Sushovan Majhi
Assistant Professor of Data Science The George Washington University Washington, D.C. s.majhi@gwu.edu
Portrait of Sushovan Majhi.
From Kawaguchiko, in the summer of MMXXV.

What is the shape of this thing? Topology's oldest question—and, posed correctly, more often the right question for data than statistics is willing to admit.

I am a mathematician: Calcutta, Bangalore, New Orleans, Berkeley, and now Washington. Each city taught a different lesson in how shape is recovered from noise. The work makes those lessons rigorous—when a Vietoris–Rips complex captures a manifold faithfully, when a Reeb graph reconstructs a road network from GPS traces, when a Gromov–Hausdorff distance is even computable. The collaborations carry those proofs into finance, climate, fluid mechanics, and biology.

Open to advising motivated students. Write.

Dispatches from the Field

From the press, 4 May 2026 — A new preprint, A Closed-Form Adaptive-Landmark Kernel for Certified Point-Cloud and Graph Classification, with Atish Mitra, Žiga Virk, and Pramita Bagchi.
From the press, 15 Mar 2026 — Interpretable Classification of Time Series Using Euler Characteristic Surfaces posted to the arXiv; under review at Nature Scientific Reports.
Newport News, 6 Mar 2025 — Spoke on Topological Stability at the Spring Topology and Dynamics Conference, Christopher Newport University.
Seattle, 9 Jan 2025 — Presented Predicting the Onset and Withdrawal of the Indian Monsoon using Persistent Homology at the Joint Mathematics Meetings.
New Orleans, 17 Nov 2024 — Returned to Tulane to speak on lower bounds for the Gromov–Hausdorff distance.
Tufts, 14 Nov 2024 — Talk at the Fall Workshop on Computational Geometry, on lower bounds for the Gromov–Hausdorff distance.
Mandi & Chennai, 10 Oct 2024 — Lectured on a taste of topological data analysis at IIT Mandi and Vellore Institute of Technology during a fortnight in India.
Athens, 14 Jun 2024 — Presented Demystifying Latschev’s Theorem for Manifold Reconstruction at SoCG 2024.
Bozeman, 28 Mar 2024 — Spoke at Montana State University on Latschev’s theorem and quantitative manifold reconstruction.
AATRN, 23 Aug 2023 — A talk at the Applied Algebraic Topology Research Network seminar on Latschev’s theorem; recording on YouTube.
No matching items

Research

Filed under Research Programme Updated MMXXVI

Two halves of the mathematical foundations of data science, with topology as the lens. On one side, theorems about when a finite sample can recover an unknown shape. On the other, the messy data that asks the question in the first place. The two halves keep each other honest.

The theoretical side develops provable methods for shape, graph, and manifold reconstruction. The objects of study are simplicial complexes built from finite samples—Vietoris–Rips, Čech, alpha—and the questions are about when, and how faithfully, such complexes recover the topology and geometry of an unknown ground truth. The tools come from algebraic topology, metric geometry, and computational geometry.

the data the topology the descriptor the classifier → → → point cloud graph Rips alpha PD d b ECS τ ε Plate I
The pipeline in four panels—data (point cloud or graph) to topology (Vietoris–Rips or alpha), to descriptor (persistence diagram or Euler characteristic surface), to a classifier in embedding space. Landmarks are ringed where they appear.

The applied side carries these methods to finance, climate, fluid mechanics, and biology—domains where data lives on or near a low-dimensional structure. Recent collaborations have addressed monsoon onset, polar vortex topology, two-phase flow regimes, and stock-market regime change.

The work, in six projects—click a card for collaborators, papers, and a longer description.

The full programme—projects, publications, and ongoing collaborations—lives on the research page.

Recent Papers

Filed under Preprints Most recent five Updated MMXXVI

The five most recent preprints from the research program. Journals, conference proceedings, and my Ph.D. thesis live on the publications page.

Each paper below is the product of a collaboration listed in the byline. Click through for the manuscript.

A Closed-Form Adaptive-Landmark Kernel for Certified Point-Cloud and Graph Classification
with Atish Mitra, Ziga Virk, Pramita Bagchi
submitted toFoundations of Computational Mathematics·2026
arxiv
Machine LearningTopological Data Analysis
A Closed-Form Persistence-Landmark Pipeline for Certified Point-Cloud and Graph Classification
with Atish Mitra, Ziga Virk, Pramita Bagchi
submitted toJournal of Machine Learning Research·2026
arxiv
Machine LearningTopological Data Analysis
Detecting Regime Transitions in Dynamical Systems via the Mixup Euler Characteristic Profile
with Atish Mitra, Santanu Nandi, Md Nurujjaman, Buddha Nath Sharma
submitted toChaos: An Interdisciplinary Journal of Nonlinear Science·2026
arxiv
Topological Data Analysis
Topological Characterization of Churn Flow and Unsupervised Correction to the Wu Flow-Regime Map in Small-Diameter Vertical Pipes
with Brady Koenig, Atish Mitra, Abigail Stein, Burt Todd
submitted toInternational Journal of Multiphase Flow (IJMF)·2026
arxiv
Topological Data AnalysisMachine LearningFluid Mechanics
Interpretable Classification of Time Series Using Euler Characteristic Surfaces
with Salam Rabindrajit Luwang, Vishal Mandal, Atish J. Mitra, Md. Nurujjaman, Buddha Nath Sharma
submitted toNature Scientific Reports·2026
arxiv
Topological Data AnalysisMachine Learning
No matching items

Students and Mentees

Filed under The Group The George Washington University 2025–26

I am fortunate to work with a talented group of students at The George Washington University.

The group reads broadly and hacks freely—topology, algorithms, statistics, optimization—and converges, eventually, on whichever piece of mathematics the data demands.

Graduate students.

  • Khush Shah, Ph.D. (Data Science), GWU
  • Shikha Kumari, Ph.D. (Data Science), GWU
  • Tyler Wallett, M.S. (Data Science), GWU

Undergraduate students.

  • James Moukheiber, GWU

I am open to advising motivated graduate and undergraduate students interested in applied topology, computational geometry, and the mathematics of machine learning. If you would like to work with me, please get in touch.

Teaching

Filed under Pedagogy GWU, Berkeley, Tulane, NIT Sikkim 2018–present

A pedagogy spanning the foundations of data science, statistics, machine learning, and topological data analysis.

Recent graduate courses at GWU include Linear Algebra for Data Science and Algorithm Design for Data Science; before that, courses at UC Berkeley (MIDS), Tulane, and NIT Sikkim. See the teaching page for the full record and the teaching statement for the philosophy behind it.

Recent Talks

Filed under Engagements Most recent four Updated MMXXVI

A selection of recent invited talks, conference presentations, and seminars. The complete record lives on the talks page.

The most recent four are listed below; the timeline runs from departmental colloquia and AMS sessions to the Symposium on Computational Geometry and the Applied Algebraic Topology Research Network seminar.

  • Mar 6, 2025

    Topological Stability
    Spring Topology and Dynamics, Christopher Newport University
    Links:

    [url] 

  • Jan 9, 2025

    Predicting the Onset and Withdrawal of the Indian Monsoon using Persistent Homology
    Joint Mathematics Meetings, Seattle
    Links:

    [url] 

    Abstract A monsoon is a wind system that seasonally reverses its direction, accompanied by corresponding changes in precipitation. The Indian monsoon is the most prominent monsoon system, primarily affecting India’s rainy season and its surrounding lands and water bodies. Every year, the onset and withdrawal of this monsoon happens sometime in May-June and September-October, respectively. Since monsoons are very complex systems governed by various weather factors with random noise, the yearly variability in the dates is significant. Despite the best efforts by the India Meteorological Department (IMD) and the South Asia Climate Outlook Forum (SCOF), forecasting the exact dates of onset and withdrawal, even within a week, is still an elusive problem in climate science. The onset and withdrawal days can be attributed to sudden changes of regime (transition to and from chaos) in the weather. During such a transition to chaos, topological signatures such as the Persistence Diagram of the system change drastically within a short period. The dynamics of the Indian monsoon depend on many weather factors such as rainfall, temperature differential, wind speed, etc. A good starting point for analyzing the weather dynamics is to use Takens’ embedding on the monsoon index to reconstruct the phase space. Then, we detect the transition to chaos using topological data analysis (TDA) by tracking the history of the death and birth of -dimensional topological features. Recently, TDA has proven successful in detecting chaos and approximating bifurcation diagrams for known dynamical systems, like the Lorenz system. In this talk, we use the historical data (1948-2015) of the Indian monsoon index to Develop an early warning system for the onset and withdrawal of the Indian monsoon. In addition to giving specific onset and withdrawal dates, the proposed warning system also produces statically significant confidence bands (widows of dates) to predict transitions for a given level of significance.
  • Nov 17, 2024

    Lower Bounding the Gromov–Hausdorff Distance
    Tulane University, New Orleans

  • Nov 14, 2024

    Lower Bounding the Gromov–Hausdorff Distance
    Fall Workshop on Computation Geometry, Tufts University

No matching items

Education

Filed under Vita 2009 – 2020

A trajectory through Calcutta, Bangalore, New Orleans, and now Washington.

  • Ph.D. in Mathematics, Tulane University, New Orleans
    2020
  • M.S. in Mathematics, Tata Institute of Fundamental Research, Bangalore
    2012
  • B.S. (Hons. in Mathematics), Ramakrishna Mission Vidyamandira, Calcutta University
    2009

𝒱ℛε(M) Set in EB Garamond and Alegreya SC. Department of Mathematics, The George Washington University. · MMXXVI ·