Hi, I’m Aryan Kumar Singh

M.Tech (Artificial Intelligence) · Research · Machine Learning

View Projects Download Resume

About Me

I am an M.Tech student in Artificial Intelligence at the Indian Institute of Science (IISc), with strong interests in machine learning, deep learning, and research-oriented system design. My work focuses on data imputation, uncertainty-aware learning, and end-to-end AI systems.

Education

M.Tech in Artificial Intelligence

Indian Institute of Science (IISc), Bangalore

Aug 2025 – Aug 2027

B.Tech in Computer Science and Engineering (CGPA: 9.37)

National Institute of Technology, Silchar

Dec 2021 – May 2025

Experience

AI Intern

Siemens Digital Industries Software (Onsite)

Jan 2025 – Jun 2025

Worked on prompt engineering for LLM-based SQL generation, data migration automation workflows, and evaluation of model effectiveness for domain-specific query generation.

Intern

Tata Consultancy Services (TCS)

May 2024 – Jul 2024

Developed a React–ROS based real-time interface for speech recognition and robot command processing, resolved image caching issues, and prepared technical documentation.

Projects

Uncertainty-Aware Decision System for Reliable ML Predictions

Designed a cost-aware ensemble ML system that estimates predictive uncertainty and defers unreliable predictions instead of forcing unsafe decisions, targeted for high-risk domains like healthcare.

Python · Scikit-learn · Uncertainty Estimation · ML Systems GitHub

Distribution Shift & Silent Failure Detection for Deployed ML Models

Built a post-deployment monitoring framework to detect covariate shift using statistical drift metrics and identify silent model failures before performance degradation becomes visible.

Python · PSI · KS Test · Model Monitoring · ML Ops GitHub

Visual Attendance Monitoring System

Facial recognition–based attendance system with automated identity verification and scalable backend architecture.

Python · OpenCV · Deep Learning GitHub

Remaining Useful Life (RUL) Prediction

Predictive modeling of turbo engine degradation to estimate remaining useful life using time-series sensor data.

Python · Time Series · Prognostics GitHub

Forest Fire Prediction

Machine learning–based system for predicting forest fire occurrence using environmental and meteorological data.

Python · ML · Environmental Data GitHub

Web Scraper

Automated web scraping tool for extracting structured information from dynamic websites.

Python · Web Scraping GitHub

Research Publications

MiGRoW: A Multi-Perspective Feature Weighting Scheme Using Gini, Entropy, and Mutual Information

Published in SipCov 2025. Introduced a multi-perspective feature weighting framework combining Gini index, entropy, and mutual information for improved learning robustness.

Edge-Guided U-Net with Frequency Domain Refinement for Efficient Skin Lesion Segmentation

Presented at the 4th IEEE BECITHCON 2025, Dhaka, Bangladesh (Paper ID: 140). Proposed an edge-aware U-Net architecture enhanced with frequency-domain refinement for accurate medical image segmentation.

ViT-SMOTE: An Attention-Guided Vision Transformer with SMOTE for Automated PCOS Diagnosis

Presented at the 4th IEEE BECITHCON 2025, Dhaka, Bangladesh (Paper ID: 138). Combined Vision Transformers with SMOTE-based imbalance handling for ultrasound-based PCOS diagnosis.

Dual-Domain Learning for Gender Classification from Ear Images using Vision Transformers and Frequency-Aware CNNs

Presented at the 13th IEEE International Conference on Intelligent Systems and Embedded Design (ISED 2025), held at National Institute of Technology Raipur, India. The work explores dual-domain feature learning by combining Vision Transformers with frequency-aware CNNs for robust ear-image-based gender classification.

A Machine Learning Based Approach to Assess Landslide Susceptibility of Assam using Novel Spatial Data

Presented at the 5th International Conference on Emerging Electronics and Automation (E2A 2025), National Institute of Technology Silchar (Paper ID: 356). Developed a spatial-data-driven machine learning framework for landslide susceptibility mapping.

Transformer-Based Self-Supervised Imputation for Medical Data

Published in ICECCME 2024, Maldives. Proposed a self-supervised transformer-based framework for handling missing values in medical datasets.

Preserving Oriya Manuscript Integrity using Transformer-Based Imputation

Published in IEEE ODICON 2024. Applied transformer-based self-supervised learning for restoring missing or degraded content in historical manuscripts.

Transformer-Based Imputation and Attention GAN Oversampling for Medical Data Processing

Published in IEEE HiPC 2024. Combined transformer-based imputation with attention-guided GAN oversampling to improve medical data analysis under class imbalance.

Research Datasets (IEEE DataPort)

Prakriti200: Ayurvedic Prakriti Assessment Dataset

A bilingual (English–Hindi) dataset consisting of responses to a standardized Prakriti assessment questionnaire based on classical Ayurvedic principles. Includes 200 respondents with 24 questionnaire items and computed dosha scores.

View Dataset on IEEE DataPort

PropDataSet: Propagandistic Fake News Dataset

A dataset of 1,499 news articles labeled as real or fake, with a subset of real articles synthetically transformed into propagandistic content using techniques such as appeal to authority, fear, and loaded language.

View Dataset on IEEE DataPort

Skills

Python C C++ HTML CSS JavaScript TensorFlow PyTorch Scikit-learn Django NumPy Pandas Git MySQL Linux Windows

Honors & Awards

Kalikrishna Mrinalini Deb Krori Gold Medal

National Institute of Technology Silchar

Awarded the Gold Medal for Best Engineering Graduate of the institute in recognition of outstanding academic performance and overall excellence.

Issued: November 2025

IEEE TCPP Travel Grant

IEEE Computer Society

Received travel grant to attend and present research work at the IEEE HiPC 2024 Conference.

2024

Creative & Personal Interests

Painting

Visual art as a medium for creative expression and conceptual exploration. Interested in abstract and nature-inspired compositions.

Poetry & Writing

Writing poetry and short reflective pieces focused on human thought, introspection, and observation.

Foosball

Actively play foosball as a recreational sport, emphasizing coordination, strategy, and competitive teamwork.

Contact