CV
My curriculum vitae.
Contact Information
| Name | Subrat Kumar Swain |
| Professional Title | PhD Researcher in Machine Learning & Security |
| qiz218247@iitd.ac.in |
Professional Summary
PhD researcher at The University of Queensland and IIT Delhi, working on adversarial machine learning and network security.
Experience
-
2020 - 2021 Bengaluru, India
Product Engineer
Cognizant Technology Solutions
- Developed and integrated a Ticket Analysis Solution into Nexa, an AutoML platform, which generates plots for volumetric & multivariate analysis and arrival patterns of issue tickets.
-
2019 - 2020 San Francisco, CA (Remote)
Machine Learning Engineer
TaiyoAI Inc.
- Automated hyper-parameter optimization of models by 80% using Bayesian optimization.
- Improved model evaluation by ranking 20+ competing machine learning models by implementing an evaluation leaderboard for various use cases.
- Improved & managed model deployment pipeline for 1000+ time series forecasting models using Apache Airflow.
-
2018 - 2020 Odisha, India
Undergraduate Research Assistant (Under Prof. Bighnaraj Naik)
VSS University of Technology
- Implemented a Deep Belief Network classifier by stacking multiple Restricted Boltzmann Machines and performed a comparative study on classification power based on Gibbs chain lengths in Gibbs sampling.
Education
-
2021 - present Brisbane, Australia & New Delhi, India
PhD
The University of Queensland - IIT Delhi
Machine Learning & Security
- Machine Learning, Meta Learning, Computer Vision, Network & System Security, Cyber-Physical Systems, Cryptography
- Supervised by Prof. Dan Kim (UQ) and Prof. Vireshwar Kumar (IIT Delhi)
-
2016 - 2020 Odisha, India
Awards
-
2026 Invited Speaker, UQ-IITD Industry Connect Workshop Series
Academia-Industry Research Partnerships in Responsible AI Across Australia and India
Invited speaker at the UQ-IITD Industry Connect Workshop Series on Responsible AI.
-
2025 ACM CCS 2025 Attendee
ACM
Attended ACM CCS 2025 at Taipei, Taiwan.
-
2024 CSRC Invited Student Speaker
The University of Queensland
Invited student speaker at the CSRC conference at UQ in 2023 and 2024.
-
2021 Prime Minister's Research Fellowship (PMRF)
Government of India
Qualified for the Prime Minister’s Research Fellows PhD fellowship in lateral entry scheme for cycle 9.
-
2021 GATE CS-IT All India Rank 800
GATE
Ranked All India 800 (top 0.8%) from 101,922 students in Graduate Aptitude Test in Engineering (GATE) - CS-IT.
-
2020 SkyDeck UC Berkeley
UC Berkeley
Selected as part of SkyDeck, UC Berkeley Start-up Acceleration program as a ML Engineer for TaiyoAI.
-
2016 JEE Advanced ~99 Percentile
IITs
Achieved ~99 percentile (All India Rank 13,187 among 12,07,058 candidates) in JEE Advanced 2016.
-
2013 National Super 100, Delhi
Director General of Police Abhyanand
Selected for National Super 100, Delhi - free residential coaching for IIT entrance examinations.
-
2009 JNVST (Navodaya) Scholar
Government of India
Qualified JNVST (Navodaya) and funded by the Govt of India to pursue free education from Standard 6 to 12.
Publications
-
2023 SPAT: Semantic-Preserving Adversarial Transformation for Perceptually Similar Adversarial Examples
European Conference on Artificial Intelligence (ECAI'23) [CORE A]
-
2024 PANDA: Practical Adversarial Attack Against Network Intrusion Detection
International Conference on Dependable Systems and Networks (DSN) [CORE A]
-
2026 ImageNet-LC: A Benchmark for Object-Centric Robustness under Localized Corruptions
International Conference on Pattern Recognition (ICPR) [CORE B]
-
2020 Deep Learning and Wavelet Transform Integrated Approach for Short-term Solar PV Power Prediction
Journal of the International Measurement Confederation [IF: 5.131]
-
2026 G2AP: Gradient-Guided Adversarial Perturbation in Network Security
IEEE European Symposium on Security and Privacy-Workshops (EuroS&P-W) [CORE A]
-
2026 GhosTurb: Practical Adversarial Perturbation Framework for IoT Networks
Submitted to AsiaCCS 2026
Skills
Languages
Interests
Projects
-
Disentangling Symbols and Movements: Factor-VAE on NAR Dataset
Applied Factor-VAE to disentangle complex primitive transformations in a few-shot visual analogical reasoning task, and demonstrated interpretability of the learned factors through latent traversal analysis.
-
Reading Noisy Captions Embedded in Images
Developed an Encoder-Decoder network using ResNet-50 for image encoding and an attention-based LSTM for extracting text embedded within images, leveraging teacher forcing for training, beam search for prediction, and BLEU score for evaluation.
References
- Prof. Dan Kim
Associate Professor, School of Electrical Engineering and Computer Science, The University of Queensland, Brisbane, Australia.
- Prof. Vireshwar Kumar
Assistant Professor, Department of Computer Science and Engineering, IIT Delhi, New Delhi, India.