About Me

I am a Final year Computer Science undergraduate (B.Tech in CSE – AI & ML) at the Institute of Engineering & Management (IEM), Kolkata. Currently preparing for GATE 2026. 🎯

My interests lie in Applied Cryptography, particularly Secure Multi Party Computation, and Machine Learning for Earth & Climate Sciences particularly with time-series data.

Looking ahead, I aim to pursue doctoral studies (Ph.D.), with the ultimate goal of becoming a professor in the future. 👨🏼‍🏫

Selected Publications

STAT-X: Short-term Atmospheric Temperature Forecasting using Machine Learning with Explainable-AI A. Sarkar, D. Guha Roy, P. Datta Theoretical and Applied Climatology, Springer, 2025 DOI: https://doi.org/10.1007/s00704-025-05693-8

AgroGenie: A Smart Approach to Agriculture Using Machine Learning A. Sarkar, K. Ganguly, S. Ghosh, D. Ghosh, S. Saha, P. Datta, D. Guha Roy Lecture Notes in Networks and Systems, Springer, 2024 DOI: https://doi.org/10.1007/978-981-97-9839-1_17

An Overview of the Discrete Logarithm Problem in Cryptography A. Sarkar, D. Guha Roy, P. Datta Lecture Notes in Networks and Systems, Springer, 2024 DOI: https://doi.org/10.1007/978-981-97-4799-3_10

Manuscripts Under Preparation

Probabilistic Short-Term Temperature Forecasting using Tuned Natural Gradient Boosting Regressor A. Sarkar, D. Guha Roy, P. Datta Manuscript under preparation.

Projects

Short-Term Temperature Prediction using Machine Learning with Explainable-AI (STAT-X)

Tech: Python, scikit-learn, NGBoost, SHAP, LIME
Repo: https://github.com/Abhiroop2004/STAT-X

Developed supervised ML pipelines for short-term minimum and maximum temperature forecasting using multivariate meteorological data.
Performed extensive feature engineering and evaluated six ML models, with Gradient Boosting Regressor and MLP achieving the best performance.
Applied explainable AI (SHAP, LIME) to identify dominant climatic drivers, improving interpretability and scientific usability.
Published in Springer (Theoretical and Applied Climatology).
Extended the framework to probabilistic forecasting using NGBoost, jointly predicting mean and uncertainty, with Bayesian hyperparameter tuning and cross-city robustness analysis.

AgroGenie: Crop Recommendation using Machine Learning

Tech: Python, scikit-learn, NumPy, Pandas
Repo: https://github.com/Abhiroop2004/AgroGenie-A-Smart-approach-to-Agriculture-using-Machine-Learning-

Led a team of five to develop a machine-learning-based crop recommendation system.
Proposed a cluster-based crop recommendation strategy, offering multiple viable crops instead of a single output.
Achieved 99.6% accuracy using Random Forest with K-fold cross-validation, combined with K-means clustering.
Presented at ICISA 2024 and awarded 2nd Best Paper.
Extended version in progress, comparing hard and soft clustering algorithms with formal cluster validation metrics.

Tech: Python, Galois, PIL, NumPy
Repo: https://github.com/Abhiroop2004/Secret-Image-Sharing

Developed a cryptographically correct Python library for:
- (k, n) Visual Cryptography,
- Lossless (k, n) and (t, k, n) Shamir’s Secret Sharing on images,
- Regional Secret Image Sharing.
Ensured exact reconstruction using finite-field arithmetic (order 256).
Ongoing work integrates image segmentation and deep-learning-based inpainting for region-aware secret sharing.
Conducted under Prof. Dr. Avishek Adhikary, Presidency University.