Research Intern

The OpportunityWe are seeking a highly driven Research Intern (Machine Learning – Materials Science) to join our scientific computing team. This role is ideal for someone who wants to work at the intersection of machine learning, deep learning, and computational materials research.You will work directly with our research engineers and scientists to design, train, and optimize ML/DL models for microstructure-based property prediction, material generation, and physics-informed AI systems. This internship offers the opportunity to contribute to foundational R&D in a fast-moving, impact-focused environment. This internship offers the opportunity to contribute to foundational R&D in a fast-moving, impact-focused environment, with a stipend of ₹15,000–₹20,000 per month and accommodation provided to support your stay during the internship.What You'll Build1. Material Property Prediction Models: You will develop ML/DL pipelines for predicting mechanical, thermal, and structural properties directly from microstructure images, simulation data, and material descriptors using CNNs, Vision Transformers, and GNNs.2. Representation Learning for Microstructures: You will architect feature extractors and embedding models to learn high-dimensional latent representations of material microstructures, enabling downstream tasks like classification, clustering, and inverse design.3. Generative Models for Material Design: You will experiment with VAEs, GANs, and Diffusion Models to generate hypothetical microstructures and explore the design space for novel materials with improved properties.4. Scientific Data & ML Infrastructure: You will help build reliable, reproducible pipelines to preprocess microscopy data, manage datasets, and execute large-scale experiments on GPUs.5. Physics-Informed and Scientific ML Systems: You will incorporate domain constraints, physics priors, and symmetry-preserving representations into ML models to ensure scientific validity and robustness.Key Responsibilities1. End-to-End Model Development: Own the full lifecycle of research features—from data preprocessing to model implementation, training, evaluation, and reporting.2. Design, Experiment, & Optimize: Implement and rigorously test deep learning architectures (CNNs, ViTs, GNNs, Diffusion Models). Conduct ablations, hyperparameter optimization, and model benchmarking.3. Ensure Reproducibility & Research Discipline: Write clean, modular, well-documented code. Use Git, experiment tracking tools, and maintain high-quality research logs.4. Drive Technical Exploration: Propose new ideas, improvements, and experimental directions across generative modelling, microstructure representation learning, and physics-informed ML.5. Collaborate & Communicate: Work closely with a multidisciplinary team of ML engineers, materials scientists, and computational researchers. Present results, insights, and technical findings clearly.Who We're Looking For- Pursuing or recently completed a Bachelor's/Master's degree in:                            - Computer Science                            - Materials Science                            - Data Science                            - Metallurgical/Mechanical Engineering                            - Applied Physics                            - Or any related computational field- Strong command of Python, PyTorch (preferred), TensorFlow- Deep learning concepts: CNNs, Transformers, Autoencoders- Scientific computing (NumPy, SciPy, Pandas)- Building and evaluating ML models on real-world datasets- Solid understanding of Optimization methods, Loss functions, regularization, Computer Vision concepts and Model evaluation metrics- Research-Oriented Mindset: Curiosity-driven, experimental, and comfortable reading research papers and implementing new ideas from literature.Highly Advantageous Qualifications- Experience with materials datasets, microstructure analysis, or FEM/simulation data- Familiarity with GNNs, Diffusion Models, or Physics-Informed Neural Networks (PINNs)- Prior work with microscopy images, dataset curation, or domain-specific preprocessing- Understanding of crystallography, phase transformations, or structure–property relationships- Knowledge of CUDA, GPU optimization, or distributed training setups