Sripadam Sujith Sai

I am Sripadam Sujith Sai, a recent Chemical Engineering graduate (Class of 2025) from NIT Rourkela, with an exchange term at IIT Hyderabad.

My research interests lie at the intersection of AI+Science and Geometric deep learning, with practical experience in computer vision, deep learning, and human–robot interaction fields I am equally passionate about.

I’ve had the opportunity to work with:

  • Prof. Kishalay Mitra at the Gokul Lab, IIT Hyderabad — worked on developing physics-informed neural networks to model complex crystallization processes in the chemical industry. Also contributed to the optimization of water distribution networks in refineries, focusing on minimizing waste and improving efficiency.
  • Dr. François Brémond at the STARS Group, INRIA (France) — research focussed on designing deep learning models capable of anticipating anomalies in spatio-temporal video sequences, enabling early detection of unusual patterns in surveillance and behavioral monitoring contexts.
  • Prof. Hatice Gunes at the AFAR Lab, University of Cambridge — contributed to research on multimodal affective computing by developing graph-based deep learning models to assess children's mental wellbeing from behavioral cues perceived by robots. These models integrated facial expressions, body posture, and voice to evaluate mental wellbeing.

Earlier, I worked on developing machine learning models for CO₂ capture and conversion.

Outside research, I enjoy playing badminton and brainstorming on creative technology applications (a few ideas are born from this).

Email  /  CV  /  Google Scholar  /  GitHub  /  LinkedIn  /  Posts

profile photo

Achievements

  • 2024 Winner, Smart India Hackathon – National innovation competition focused on real-world problem-solving

Publications and Preprints

Hybrid QLSTM Paper Image QGAPHEnsemble: Combining Hybrid QLSTM Network Ensemble via Adaptive Weighting for Short Term Weather Forecasting.
Sen, A. et al. (2025). A Survey of Hyperparameter Selection Methods for Weather Forecasting Using State-of-the-Art Machine Learning Algorithms. In: Bäck, T., et al. Computational Intelligence. IJCCI 2023. Studies in Computational Intelligence, vol 1196. Springer, Cham.
https://doi.org/10.1007/978-3-031-85252-7_15
IJCCI 2023
arXiv | Springer Nature
QLSTM Paper Image QGAPHnet: Quantum Genetic Algorithm Based Hybrid QLSTM Model for Soil Moisture Estimation.
S. Sai et al., "QGAPHnet : Quantum Genetic Algorithm Based Hybrid QLSTM Model for Soil Moisture Estimation," IGARSS 2024 - 2024 IEEE International Geoscience and Remote Sensing Symposium, Athens, Greece, 2024, pp. 5191-5194, doi: 10.1109/IGARSS53475.2024.10641651.
IEEE IGARSS 2024
arXiv | IEEE Xplore
HBO-DEViT Paper Image HBO-DEViT: Vision Transformer Based Attention-Guided Evolutionary Architecture for Ship-Iceberg Categorisation in Arctic SAR Images.
A. Sen, S. Sai, C. Mallick, S. Roy and U. Sen, "HBO-DEViT: Vision Transformer Based Attention-Guided Evolutionary Architecture for Ship-Iceberg Categorisation in Arctic Sar Images," IGARSS 2024 - 2024 IEEE International Geoscience and Remote Sensing Symposium, Athens, Greece, 2024, pp. 201-204, doi: 10.1109/IGARSS53475.2024.10641319.
IEEE IGARSS 2024
arXiv | IEEE Xplore

Projects

Human-Robot Interaction Project ChesSAI : A Hybrid Reinforcement Learning-Based Self-Evolving Chess Engine
A chess engine combining PPO and MCTS, trained via self-play, imitation learning, and dynamic ELO-based curriculum.
GitHub