About

Sameer Singh is a Professor of Computer Science at UC Irvine. His primary research focuses on the robustness and interpretability of machine learning algorithms and models that reason with text and structure for natural language processing. He has worked as a postdoctoral researcher at the University of Washington and earned his Ph.D. from the University of Massachusetts, Amherst. Dr. Singh has been recognized with several awards, including being named the Kavli Fellow by the National Academy of Sciences, receiving the NSF CAREER award, the UCI Distinguished Early Career Faculty award, the Hellman Faculty Fellowship, and being selected as a DARPA Riser. His research group has received funding from notable organizations such as the Allen Institute for AI, Amazon, NSF, DARPA, Adobe Research, Hasso Plattner Institute, NEC, Base 11, and FICO. He has published extensively in machine learning and natural language processing venues and has received conference paper awards at KDD 2016, ACL 2018, EMNLP 2019, AKBC 2020, ACL 2020, and NAACL 2022.

Research topics

• Computer Science
• Artificial Intelligence
• Natural Language Processing
• Machine Learning
• Programming language
• Data Mining
• Geology
• Computer network
• Distributed computing
• Human–computer interaction
• Engineering
• History
• Archaeology

Selected publications

• Software based medical device evaluation through reinforcement learning

  AIP conference proceedings · 2026-01-01

  article
  PublisherDOI

• Waste Management System ECOSORT -AI

  INTERANTIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING AND MANAGEMENT · 2025-01-13 · 3 citations

  articleOpen accessSenior author

  Abstract- Efficient waste management is crucial for sustainable urban living. However, challenges such as improper segregation, low recycling rates, and reliance on manual systems hinder progress toward environmental goals. This paper introduces EcoSort AI, an AI-driven waste management solution that combines computer vision, IoT technologies, and machine learning to automate and optimize waste segregation. Leveraging convolutional neural networks (CNNs), the system identifies, classifies, and sorts waste materials into appropriate categories, ensuring improved recycling rates and reduced landfill burden. EcoSort AI features IoT-enabled smart bins for real-time classification and integrates seamlessly into existing urban infrastructures. Experimental results demonstrate significant improvements in sorting accuracy, efficiency, and public engagement. Index Terms- Waste segregation, artificial intelligence, IoT-enabled bins, CNNs, smart cities, recycling optimization, sustainable development.

  PublisherOA PDFDOI

• Entomotoxic efficacy of nanoparticles and elevated levels of CO2 against groundnut bruchid, Caryedon serratus (Olivier, 1790) (Coleoptera: Chrysomelidae) in stored groundnut

  Journal of Stored Products Research · 2025-07-25

  article
  PublisherDOI

• Chemical Protein Engineering: Backbone Cyclization Rescues Folding of a 183‐Residue Truncated Domain of Malaria Parasite Protein <i>Pf</i>AMA1

  Chemistry - A European Journal · 2025-04-08

  articleOpen access

  The interaction between apical membrane antigen 1 (PfAMA1) and rhoptry neck protein 2 (PfRON2) is crucial for Plasmodium falciparum red blood cell invasion, making it a key target for anti-malarial drug development strategies. Here, we report the chemical synthesis of PfAMA1 domain I (PfAMA1-DI) in both linear and backbone-circularized forms, employing a six-segment convergent synthesis approach exploiting one-pot chemistries and solubilizing tags. The chemically synthesized linear PfAMA1-DI construct exhibited incomplete disulfide bond formation during folding, likely due to increased terminal flexibility in the absence of domain II. To address this, we employed backbone cyclization of the large 180-residue polypeptide chain, with 3-residue linker sequence, as a unique strategy to conformationally restrict its termini and facilitate correct disulfide bond formation. Introducing a multipurpose affinity and solubility tag to the cyclicPfAMA1-DI construct further improved the folding yield by mitigating aggregation. The predicted structure using ColabFold-Alphafold2 indicated that PfRON2 ligand binds within the hydrophobic groove of the cyclicPfAMA1-DI construct similar to the native interactions. These findings underscore the potential of large protein backbone cyclization to stabilize protein structure, offering a compelling strategy for the chemical synthesis of otherwise unstable protein domains with broad applications in miniature protein engineering.

  PublisherOA PDFDOI

• Transformation of Vehicular Networks through Machine Learning: Challenges and Opportunities

  2025-06-16

  article1st authorCorresponding

  Autonomous vehicles (AVs) are currently getting widespread attention and considered as a highly promising application of wireless communication. Due to the standout features of 5G/6G, it is capable of supporting V2X (Vehicular to Everything) communications and can fulfill the communication requirements. For the efficient vehicular ad-hoc network (VANET), lower latency and ultra-reliability are prime requirements. Next, larger handoffs, high interference, and dynamic traffic are the major obstacles to seamless connectivity. These issues can be effectively tackled through the concept of Machine Learning (ML). In this paper, we conduct a comprehensive survey of vehicular communication, discussing its major challenges, highlighting the transformative potential of ML algorithms, and addressing the implementation challenges of ML for creating a smooth vehicular network. Furthermore, we explore future research opportunities in this direction.

  PublisherDOI

• RF FSO integration for hand-off reduction in CRAN assisted vehicular networks

  Journal of Optics · 2025-05-17 · 1 citations

  article
  PublisherDOI

• Clinical and Experimental Studies of Structural Valve Degeneration of Bovine Jugular Vein Valves: Mitigation with Polyoxazoline Modification

  Pediatric Cardiology · 2025-10-08

  articleOpen access
  PublisherOA PDFDOI

• Characterizing Mamba’s Selective Memory using Auto-Encoders

  2025-01-01

  articleOpen access

  Tamanna Hossain, Robert L. Logan Iv, Chandrasekhara Ganesh Jagadeesan, Sameer Singh, Joel R. Tetreault, Alejandro Jaimes. Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics. 2025.

  PublisherOA PDFDOI

• TurtleBench: A Visual Programming Benchmark in Turtle Geometry

  2025-01-01

  articleOpen access

  Sina Rismanchian, Yasaman Razeghi, Sameer Singh, Shayan Doroudi. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

  PublisherOA PDFDOI

• Semantic Probabilistic Control of Language Models

  ArXiv.org · 2025-05-04

  preprintOpen accessSenior author

  Semantic control entails steering LM generations towards satisfying subtle non-lexical constraints, e.g., toxicity, sentiment, or politeness, attributes that can be captured by a sequence-level verifier. It can thus be viewed as sampling from the LM distribution conditioned on the target attribute, a computationally intractable problem due to the non-decomposable nature of the verifier. Existing approaches to LM control either only deal with syntactic constraints which cannot capture the aforementioned attributes, or rely on sampling to explore the conditional LM distribution, an ineffective estimator for low-probability events. In this work, we leverage a verifier's gradient information to efficiently reason over all generations that satisfy the target attribute, enabling precise steering of LM generations by reweighing the next-token distribution. Starting from an initial sample, we create a local LM distribution favoring semantically similar sentences. This approximation enables the tractable computation of an expected sentence embedding. We use this expected embedding, informed by the verifier's evaluation at the initial sample, to estimate the probability of satisfying the constraint, which directly informs the update to the next-token distribution. We evaluated the effectiveness of our approach in controlling the toxicity, sentiment, and topic-adherence of LMs yielding generations satisfying the constraint with high probability (&gt;95%) without degrading their quality.

  PublisherOA PDFDOI

Recent grants

• Collaborative Research: RI: Small: Post hoc Explanations in the Wild: Exposing Vulnerabilities and Ensuring Robustness

  NSF · $225k · 2020–2023

• RI: Small: Modeling Multiple Modalities for Knowledge-Base Construction

  NSF · $448k · 2018–2022

• CRII: RI: Explaining Decisions of Black-box Models via Input Perturbations

  NSF · $175k · 2018–2021

Frequent coauthors

• Matt Gardner

  Duke Institute for Health Innovation

  110 shared

• Robert L. Logan

  61 shared

• Eric Wallace

  49 shared

• Dheeru Dua

  30 shared

• Nitish Gupta

  National Institute of Technology Warangal

  30 shared

• Sebastian Riedel

  29 shared

• Pouya Pezeshkpour

  27 shared

• Dylan Slack

  27 shared

Education

• PhD, Computer Science

  University of Massachusetts Amherst

  2014

• MS, EECS

  Vanderbilt University

  2007

Awards & honors

• Kavli Fellow by the National Academy of Sciences
• NSF CAREER award
• UCI Distinguished Early Career Faculty award
• Hellman Faculty Fellowship
• Selected as a DARPA Riser