Rupam Kumawat

Researcher

Delhi, India1 yr 3 mos experience

AI EnabledAI ML Practitioner

Key Highlights

Expert in Explainable AI and Graph Machine Learning.
Developed scalable methods for GNN explainability.
Created interactive chatbots integrating advanced AI technologies.

Stackforce AI infers this person is a specialist in Explainable AI and Graph Machine Learning within the Fintech and Agritech sectors.

Contact

Skills

Core Skills

Graph Neural NetworksExplainable AiConvolutional Neural Networks (cnn)Long Short-term Memory (lstm)Artificial Intelligence (ai)Chatbot Development

Other Skills

Research SkillsScalabilityMachine LearningSatellite ImagerySemi supervised learningKMeansAlgorithm DevelopmentPrompt EngineeringAPI DevelopmentSpeech RecognitionResearch and Development (R&D)Text-to-Speech SynthesisPython (Programming Language)Large Language Models (LLM)Generative AI

About

I’m deeply interested in the question “How can we make deep learning models explainable, transparent and interpretable?” My research interests lie in Explainable AI, Graph Machine Learning, and trustworthy, scalable AI systems. I’m passionate about developing methods that make complex models more transparent, reliable, and mathematically grounded, bridging theory with real-world machine learning applications. I’m always eager to explore ideas that push the boundaries of interpretable and responsible AI.

Experience

Nanyang technological university singapore

GCF Fellow

Feb 2026 – Present · 1 mo · Singapore, Singapore · On-site

Indian institute of technology, delhi

2 roles

Teaching Assistant

Jan 2026 – Present · 2 mos · On-site

Teaching Assistant

Jul 2025 – Nov 2025 · 4 mos · New Delhi, Delhi, India · On-site

Teaching Assistant of an undergraduate-level course, Numerical Methods and Analysis

Inter iit tech meet 14.0

Participant- Inter IIT Tech 14.0

Sep 2025 – Dec 2025 · 3 mos

Machine intelligence signal and network (misn lab), iit delhi

Master Thesis Student

Jul 2025 – Present · 8 mos · Delhi, India · On-site

Graphs are pervasive in a wide range of domains, from social networks and biological systems to knowledge graphs and communication infrastructures. However, handling large-scale graphs presents substantial computational challenges, and their inherently evolving nature further complicates analysis. My research focuses on online graph coarsening, a framework designed to address these two fundamental issues—scalability and adaptability to dynamic graph evolution. By developing efficient, incremental coarsening techniques, my work aims to enable real-time processing and learning over continuously changing large-scale graph data.
Supervised by: Dr. Sandeep Kumar

Mastercard

ML Research Intern

May 2025 – Jul 2025 · 2 mos · Gurugram, Haryana, India · On-site

Most deep learning models, including Graph Neural Networks (GNNs), operate as black boxes, making their decision processes difficult to interpret. My research focused on developing a scalable GNN explainability method capable of interpreting the decisions made by the in-production GNN model at Mastercard. The proposed approach was specifically designed to handle the large-scale transaction network, achieving a 10× improvement in computational efficiency compared to existing methods. This work not only resulted in a patent application but also laid the foundation for a new research direction that leverages explainability to enhance GNN performance.

Graph Neural NetworksExplainable AIResearch SkillsScalabilityMachine Learning

University of minnesota

Summer Research Intern

May 2024 – Aug 2024 · 3 mos · Minneapolis, Minnesota, United States · On-site

Crop detection is an active field of research, and it helps the authorities to analyse the crop growing patterns across the region and facilitates them to make more informed decisions. Cover crops are a type of crop that helps to enrich soil nutrition. My work was mainly focused on developing a deep learning model that can be used for cover crop detection using satellite imagery of a region. Our team proposed to use a CNN-LSTM architecture to capture the spatio-temporal nature of the data, and due to a lack of ground truth data, we also proposed a method for label propagation using K-Means Clustering.
Supervised by: Dr. Vipin Kumar

Convolutional Neural Networks (CNN)Long Short-term Memory (LSTM)Satellite ImagerySemi supervised learningKMeans

Gopinnacle ai

AI/ML Intern

Nov 2023 – Jan 2024 · 2 mos · India · Remote

The primary goal of the startup was to create an interactive platform for real-life interview preparation. Firstly, I worked on building speech-to-text and text-to-speech algorithms. Later, as per business requirements, I worked on RAG and LLMs to generate and evaluate answers provided by the user.

Aiscf-iit delhi

Student Fellow

Aug 2023 – Oct 2023 · 2 mos · New Delhi, Delhi, India · On-site

I was selected as a student fellow, and I deep-dived into the concepts of AI Safety and the future of AI. I also explored the aspects of explainable and interpretable AI.

Algorithm DevelopmentPrompt EngineeringAPI DevelopmentSpeech RecognitionResearch and Development (R&D)Text-to-Speech Synthesis

Cardiff university / prifysgol caerdydd

Research Intern

Jun 2023 – Jul 2023 · 1 mo · London, England, United Kingdom · Remote

I implemented an interactive chattable virtual avatar by integrating three GitHub repositories—LangChain, Bark, and SadTalker. For the workflow, I used Cohere as the chat model, LangChain as the chatbot, Bark for audio generation, and SadTalker to animate a talking head. The project successfully combined text, audio, and visual components into a seamless conversational avatar. My supervisor recognized my contributions and awarded me a Letter of Recommendation for my valuable and constructive input.
Supervised by: Dr. Yipeng Qin

Artificial Intelligence (AI)