About

Aaditya Ramdas is an Associate Professor (with tenure) at Carnegie Mellon University in the Department of Statistics and Data Science and the Machine Learning Department, with a planned move to Stanford University starting September 1, 2026. His academic background includes a PhD from CMU, where he was awarded the Umesh K. Gavaskar Memorial Thesis Award, and an undergraduate degree in Computer Science from IIT Bombay, where he received a Young Alumnus Achiever Award in 2026. His postdoctoral work was conducted at UC Berkeley under the mentorship of Michael Jordan and Martin Wainwright. His research focuses on mathematical statistics and learning, with an emphasis on designing algorithms that combine strong theoretical guarantees with practical effectiveness. His main interests include post-selection inference, game-theoretic statistics such as e-values and confidence sequences, and predictive uncertainty quantification through conformal prediction and calibration. He applies his work to areas like privacy, neuroscience, genetics, and auditing. Ramdas has published over 150 peer-reviewed papers in top journals and AI conferences, and has given keynote talks and tutorials at major venues. He has received numerous awards, including the Presidential Early Career Award (PECASE), Sloan Fellowship, Kavli Fellowship, and was elected Fellow of the IMS. He has also served as program chair and general chair for AISTATS and has been involved in organizing workshops on game-theoretic statistics and sequential inference. Additionally, he co-organizes the StatML Group at CMU and maintains a list of companies utilizing his inference methods in deployed software.

Research topics

• Algorithm
• Applied mathematics
• Mathematical optimization
• Statistics
• Combinatorics
• Mathematics

Selected publications

• Post-detection inference for sequential changepoint localization

  Journal of the Royal Statistical Society Series B (Statistical Methodology) · 2026-04-08

  preprintOpen accessSenior author

  Abstract This article addresses a fundamental but largely unexplored challenge in sequential changepoint analysis: conducting inference following a detected change. We develop a very general framework to construct confidence sets for the unknown changepoint using only the data observed up to a data-dependent stopping time at which an arbitrary sequential detection algorithm declares a change. Our framework is nonparametric, making no assumption on the composite postchange class, the observation space, or the sequential detection procedure used, and is nonasymptotically valid. We also extend it to handle composite prechange classes under a suitable assumption and also derive confidence sets for the change magnitude in parametric settings. We provide theoretical guarantees on the width of our confidence intervals. Extensive simulations demonstrate that the produced sets have reasonable size, and slightly conservative coverage. In summary, we present the first general method for sequential changepoint localization, which is theoretically sound and broadly applicable in practice.

  PublisherOA PDFDOI

• Universal Sequential Changepoint Detection of Quantum Observables via Classical Shadows

  Open MIND · 2026-02-12

  preprintSenior author

  We study sequential quantum changepoint detection in settings where the pre- and post-change regimes are specified through constraints on the expectation values of a finite set of observables. We consider an architecture with separate measurement and detection modules, and assume that the observables relevant to the detector are unknown to the measurement device. For this scenario, we introduce shadow-based sequential changepoint e-detection (eSCD), a novel protocol that combines a universal measurement strategy based on classical shadows with a nonparametric sequential test built on e-detectors. Classical shadows provide universality with respect to the detector's choice of observables, while the e-detector framework enables explicit control of the average run length (ARL) to false alarm. Under an ARL constraint, we establish finite-sample guarantees on the worst-case expected detection delay of eSCD. Numerical experiments validate the theory and demonstrate that eSCD achieves performance competitive with observable-specific measurement strategies, while retaining full measurement universality.

  DOI

• An Asymptotic Law of the Iterated Logarithm for $\mathrm{KL}_{\inf}$

  Open MIND · 2026-02-05

  preprintSenior author

  The population $\mathrm{KL}_{\inf}$ is a fundamental quantity that appears in lower bounds for (asymptotically) optimal regret of pure-exploration stochastic bandit algorithms, and optimal stopping time of sequential tests. Motivated by this, an empirical $\mathrm{KL}_{\inf}$ statistic is frequently used in the design of (asymptotically) optimal bandit algorithms and sequential tests. While nonasymptotic concentration bounds for the empirical $\mathrm{KL}_{\inf}$ have been developed, their optimality in terms of constants and rates is questionable, and their generality is limited (usually to bounded observations). The fundamental limits of nonasymptotic concentration are often described by the asymptotic fluctuations of the statistics. With that motivation, this paper presents a tight (upper and lower) law of the iterated logarithm for empirical $\mathrm{KL}_{\inf}$ applying to extremely general (unbounded) data.

  DOI

• Universal Sequential Changepoint Detection of Quantum Observables via Classical Shadows

  ArXiv.org · 2026-02-12

  articleOpen accessSenior author

  We study sequential quantum changepoint detection in settings where the pre- and post-change regimes are specified through constraints on the expectation values of a finite set of observables. We consider an architecture with separate measurement and detection modules, and assume that the observables relevant to the detector are unknown to the measurement device. For this scenario, we introduce shadow-based sequential changepoint e-detection (eSCD), a novel protocol that combines a universal measurement strategy based on classical shadows with a nonparametric sequential test built on e-detectors. Classical shadows provide universality with respect to the detector's choice of observables, while the e-detector framework enables explicit control of the average run length (ARL) to false alarm. Under an ARL constraint, we establish finite-sample guarantees on the worst-case expected detection delay of eSCD. Numerical experiments validate the theory and demonstrate that eSCD achieves performance competitive with observable-specific measurement strategies, while retaining full measurement universality.

  PublisherOA PDF

• Improving Wald’s (Approximate) Sequential Probability Ratio Test by Avoiding Overshoot

  IEEE Transactions on Information Theory · 2026-01-28

  articleOpen accessSenior author

  Wald’s sequential probability ratio test (SPRT) is a cornerstone of sequential analysis. Based on desired type-I, II error levels α, β, it stops when the likelihood ratio crosses certain thresholds, guaranteeing optimality of the expected sample size. However, these thresholds are not closed form and the test is often applied with approximate thresholds (1 – β)/α and β/(1 – α) (approximate SPRT). When β > 0, this neither guarantees error control at α, β nor optimality. When β = 0 (power-one SPRT), this method is conservative and not optimal. The looseness in both cases is caused by <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">overshoot</i>: the test statistic overshoots the thresholds at the stopping time. Numerically calculating thresholds may be infeasible, and most software packages do not do this. We improve the approximate SPRT by modifying the test statistic to avoid overshoot. Our ‘sequential boosting’ technique <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">uniformly</i> improves power-one SPRTs (β = 0) for simple nulls and alternatives, or for one-sided nulls and alternatives in exponential families. When (β > 0), our techniques provide guaranteed error control at α, β, while needing less samples than the approximate SPRT in our simulations. We also provide several nontrivial extensions: confidence sequences, sampling without replacement and conformal martingales.

  PublisherDOI

• Operationalizing Stein's Method for Online Linear Optimization: CLT-Based Optimal Tradeoffs

  arXiv (Cornell University) · 2026-02-06

  articleOpen accessSenior author

  Adversarial online linear optimization (OLO) is essentially about making performance tradeoffs with respect to the unknown difficulty of the adversary. In the setting of one-dimensional fixed-time OLO on a bounded domain, it has been observed since Cover (1966) that achievable tradeoffs are governed by probabilistic inequalities, and these descriptive results can be converted into algorithms via dynamic programming, which, however, is not computationally efficient. We address this limitation by showing that Stein's method, a classical framework underlying the proofs of probabilistic limit theorems, can be operationalized as computationally efficient OLO algorithms. The associated regret and total loss upper bounds are "additively sharp", meaning that they surpass the conventional big-O optimality and match normal-approximation-based lower bounds by additive lower order terms. Our construction is inspired by the remarkably clean proof of a Wasserstein martingale central limit theorem (CLT) due to Röllin (2018). Several concrete benefits can be obtained from this general technique. First, with the same computational complexity, the proposed algorithm improves upon the total loss upper bounds of online gradient descent (OGD) and multiplicative weight update (MWU). Second, our algorithm can realize a continuum of optimal two-point tradeoffs between the total loss and the maximum regret over comparators, improving upon prior works in parameter-free online learning. Third, by allowing the adversary to randomize on an unbounded support, we achieve sharp in-expectation performance guarantees for OLO with noisy feedback.

  PublisherOA PDF

• Asymptotically optimal sequential change detection for bounded means

  Open MIND · 2026-02-05

  preprintSenior author

  We consider the problem of quickest changepoint detection under the Average Run Length (ARL) constraint where the pre-change and post-change laws lie in composite families $\mathscr{P}$ and $\mathscr{Q}$ respectively. In such a problem, a massive challenge is characterizing the best possible detection delay when the "hardest" pre-change law in $\mathscr{P}$ depends on the unknown post-change law $Q\in\mathscr{Q}$. And typical simple-hypothesis likelihood-ratio arguments for Page-CUSUM and Shiryaev-Roberts do not at all apply here. To that end, we derive a universal sharp lower bound in full generality for any ARL-calibrated changepoint detector in the low type-I error ($γ\to\infty$ regime) of the order $\log(γ)/\mathrm{KL}_{\mathrm{inf}}(Q,\mathscr{P})$. We show achievability of this universal lower bound by proving a tight matching upper bound (with the same sharp $\logγ$ constant) in the important bounded mean detection setting. In addition, for separated mean shifts, we also we derive a uniform minimax guarantee of this achievability over the alternatives.

  DOI

• Multiple testing with anytime-valid Monte Carlo p-values

  Electronic Journal of Statistics · 2026-01-01

  articleOpen accessSenior author

  In contemporary problems involving genetics or neuroimaging data, thousands of hypotheses need to be tested. Due to their high power, and finite sample guarantees on type-I error under weak assumptions, Monte Carlo permutation tests are often considered as gold standard for these settings. However, the enormous computational effort required for (thousands of) permutation tests is a major burden. In this paper, we integrate recently constructed anytime-valid permutation p-values into a broad class of multiple testing procedures, including the Benjamini-Hochberg procedure. This allows to fully adapt the number of permutations to the underlying data and thus, for example, to the number of rejections made by the multiple testing procedure. Even though this data-adaptive stopping can induce dependencies between the p-values that violate the usual assumptions of the Benjamini-Hochberg procedure, we prove that our approach controls the false discovery rate under mild assumptions. Furthermore, our method provably decreases the required number of permutations substantially without compromising power. On a real RNA sequencing genomics dataset, our method reduced the computational time from over two and a half days to under two minutes while increasing the number of rejections.

  PublisherDOI

• On admissibility in post-hoc hypothesis testing

  International Journal of Approximate Reasoning · 2026-01-21

  articleOpen access
  PublisherDOI

• Conformal changepoint localization

  Open MIND · 2026-02-05

  preprintSenior author

  We study the problem of offline changepoint localization in a distribution-free setting. One observes a vector of data with a single changepoint, assuming that the data before and after the changepoint are iid (or more generally exchangeable) from arbitrary and unknown distributions. The goal is to produce a finite-sample confidence set for the index at which the change occurs without making any other assumptions. Existing methods often rely on parametric assumptions, tail conditions, or asymptotic approximations, or only produce point estimates. In contrast, our distribution-free algorithm, CONformal CHangepoint localization (CONCH), only leverages exchangeability arguments to construct confidence sets with finite sample coverage. By proving a conformal Neyman-Pearson lemma, we derive principled score functions that yield informative (small) sets. Moreover, with such score functions, the normalized length of the confidence set shrinks to zero under weak assumptions. We also establish a universality result showing that any distribution-free changepoint localization method must be an instance of CONCH. Experiments suggest that CONCH delivers precise confidence sets even in challenging settings involving images or text.

  DOI

Recent grants

• Nonparametric Confidence Sequences and their Applications

  NSF · $160k · 2019–2022

• CAREER: Online Multiple Hypothesis Testing: A Comprehensive Treatment

  NSF · $400k · 2020–2025

• Game-theoretic statistics and safe anytime-valid inference

  NSF · $160k · 2023–2026

Frequent coauthors

• Michael I. Jordan

  37 shared

• Larry Wasserman

  Carnegie Mellon University

  35 shared

• Martin J. Wainwright

  Massachusetts Institute of Technology

  32 shared

• Rina Foygel Barber

  23 shared

• Aarti Singh

  22 shared

• Ryan J. Tibshirani

  20 shared

• Ian Waudby-Smith

  Carnegie Mellon University

  19 shared

• Emmanuel J. Candès

  15 shared

Education

• B.S., Computer Science

  IIT Bombay

  2009

• Ph.D.

  Carnegie Mellon University

  2015

Awards & honors

• Umesh K. Gavaskar Memorial Thesis Award
• Young Alumnus Achiever Award (2026)
• Presidential Early Career Award (PECASE)
• Kavli Fellowship from the National Academy of Sciences
• Sloan Fellowship in Mathematics