CV
This is a description of the page. You can modify it in '_pages/cv.md'. You can also change or remove the top pdf download button.
Contact Information
| Name | Deepak Kumar |
| Professional Title | Materials Researcher & Data Scientist |
| deepak72855@gmail.com | |
| Phone | +91-8439373413 |
| Location | New Delhi, Delhi |
| Website | https://deep7285.github.io |
Professional Summary
Materials researcher and data scientist working at the intersection of materials characterisation, computer vision, and machine learning. MS by Research from IIT Madras (2025) with a thesis on friction stir welding of dissimilar materials. Currently a Data Scientist at IIT Delhi, applying physics-informed ML to high-dimensional spatio-temporal ocean data. Open-source contributor and active collaborator with Oxford Nano Analysis Group, University of Oxford.
Experience
-
August - New Delhi, India
Data Scientist
Indian Institute of Technology Delhi
Physics-informed ML and deep learning on high-dimensional spatio-temporal ocean data for extreme marine heatwave event detection and forecasting.
- Designing ConvLSTM and temporal CNN architectures for spatiotemporal modelling on gridded ocean reanalysis data
- Physics-informed feature engineering — encoding domain knowledge of mixed layer dynamics and air-sea heat flux anomalies as model constraints
- Statistical characterisation of extreme event distributions before predictive modelling
- Evaluating model outputs against known physical drivers to ensure physical consistency alongside statistical accuracy
-
February - February Hyderabad, India
Associate Solution Leader — Machine Learning Engineer
Brane Enterprises Pvt. Ltd.
ML engineering within a 50-person team, building deep learning solutions for e-commerce and retail clients.
- Developed deep learning models (TensorFlow, PyTorch) for product price optimisation and customer segmentation — contributed to 15% increase in sales for e-commerce clients
- Built analytics pipelines using Tableau and Apache Superset — 10% improvement in sales forecast accuracy for retail clients
- Collaborated on model evaluation, feature engineering, and deployment workflows alongside senior engineers
-
February - June Chennai, India
Teaching Assistant
IIT Madras — MEMS Lab, Dept. of Metallurgical & Materials Engineering
Teaching Assistant for two courses under Prof. Ranjit Bauri.
- MM2010 Principles of Physical Metallurgy (Aug 2022 – Nov 2023): led weekly lab sessions for a batch of 70 students
- NOC24-MM14 Theory and Practice of Non-Destructive Testing (Dec 2023 – Mar 2024): assisted in course delivery and examinations
-
October - December Uttar Pradesh, India
Technical Consultant
OSN Consulting & Associates
Technical and financial valuation of large-scale plant and machinery projects across manufacturing and infrastructure sectors.
- Conducted technical and financial valuations (INR 150–200 Cr range) supporting strategic procurement and investment decisions
- Developed rigorous cross-functional technical reports bridging engineering and finance teams
-
July - September Kota, Rajasthan, India
Assistant Engineer
Dynamic Intra-tech Pvt. Ltd.
Production management in steel bridge fabrication.
- Managed production team in steel bridge fabrication; optimised machining parameters through data-driven analysis of production records
- Achieved 9–10% improvement in production throughput
-
February - May Chennai, India
Student Member
IIT Madras — CDC-R (Career Development Cell, Research)
Supported research scholars at IIT Madras with career planning, industry connections, and professional development initiatives.
-
- India
Delegate — Country of Niger
35th IMUN Conference
Represented the country of Niger on the topic of gender equality at the International Model United Nations conference.
Education
-
February - July Chennai, India
MS by Research
Indian Institute of Technology Madras
Metallurgical and Materials Engineering
- Texture in Materials
- Modern Techniques of Material Characterization
- X-Ray Diffraction Techniques
- Physical Metallurgy
- Mechanical Behaviour of Materials
- Advanced Welding Techniques
- Aluminium Alloys and their Composites
- Sheet Metal Forming
- Thesis: Friction Stir Welding of Dissimilar AA6082-T6 and DP780 Steel Joints for Lightweight Automotive Applications — Guide: Prof. Ranjit Bauri
- Presented at THERMEC’23 International Conference, University of Technology Vienna, Austria
- 3rd place — Amalgam National Symposium, IIT Madras (2024)
-
August - May Uttar Pradesh, India
Bachelor of Technology
BBDNITM — Dr. APJ Abdul Kalam Technical University
Mechanical Engineering
- Final year project: Automatic Kickstand Retrieval System in Bike to Prevent Accidents — used AISI 4130 Steel and Titanium Grade 5 alloy
Awards
-
February GATE Scholarship — 94.2 Percentile
Government of India
Awarded GATE (Graduate Aptitude Test in Engineering) scholarship with 94.2 percentile score in Mechanical Engineering, enabling postgraduate study at IIT Madras.
-
2024 3rd Place — Research Presentation, Amalgam National Symposium
IIT Madras
Secured 3rd place in research presentation at the Amalgam National Symposium 2, IIT Madras.
-
2023 Top 10% — Google Analytics Workshop
IIT Madras
Finished among the top 10% of participants at the Google Analytics workshop held at IIT Madras.
Publications
-
2026 Microstructure and Mechanical Properties of Friction Stir Welded AA6082 Aluminium Alloy and Dual-Phase 780 Steel Dissimilar Joints
Journal of Manufacturing Processes
1st author · Under review. Investigates FSW of AA6082-T6 and DP780 steel lap joints; correlates process parameters with microstructure, IMC formation, and mechanical performance using SEM, EBSD, XRD, and nano-indentation.
-
2025 Microstructural Evolution and Corrosion Kinetics of X2 Zr-Nb Alloy in Simulated PWR and BWR Environments
ASTM STP Proceedings
2nd author · Accepted for publication. Studies dissolved oxygen effects on corrosion of Zr-Nb nuclear cladding in simulated reactor environments using TEM and spectroscopy. Collaboration with Oxford Nano Analysis Group.
-
2025 PoroDet: A Python Package for Nano-porosity Detection in TEM Images
Journal of Open Source Software (JOSS)
Co-author · Submitted. Open-source Python package for automated segmentation of nano-porosities in Fresnel-contrast TEM images using U-Net CNN. Developed with Oxford Nano Analysis Group.
-
2025 Comparative Performance of Classical and Deep Learning Architectures for Nano-porosity Detection in TEM Images
1st author · Under preparation. Benchmarks Random Forest, U-Net, and Vision Transformer (Trans-UNet) approaches for nano-porosity segmentation; validates with ROC-AUC, precision-recall, and confusion matrices.
Skills
Languages
Interests
Certificates
- THERMEC'23 International Conference — Oral Presentation - University of Technology Vienna, Austria (2023)
- SPARC International Workshop on Aluminum Alloys - IIT Madras & McMaster University (2023)
- Symbiosis AI/ML Conference and Workshop - IIT Madras (2023)
Projects
-
PoroDet
Open-source Python package for automated nano-porosity segmentation in TEM images using U-Net CNN. Submitted to JOSS; published on Zenodo.
- U-Net architecture achieving 92% feature detection accuracy
- Designed for limited training data using augmentation
- JOSS submission · Zenodo DOI: 10.5281/zenodo.19342741
-
Physics-Informed Surrogate Modelling for Dissimilar FSW
GPR surrogate model with Bayesian optimisation for the process–structure–property design space of dissimilar friction stir welding.
- Gaussian Process Regression with uncertainty quantification
- Bayesian optimisation for intelligent experiment selection
- Physics-informed process–structure–property chain
-
Marine Heatwave Detection & Forecasting
Physics-aware deep learning for spatiotemporal detection and forecasting of extreme marine heatwave events in the North Indian Ocean.
- ConvLSTM and temporal CNN architectures
- Physics-informed feature engineering from ocean reanalysis data
- Spatiotemporal modelling of extreme event distributions
-
Human Mentor–Mentee Matching System
NLP-based alumni mentor–mentee matching pipeline for IIT Madras alumni network. Live deployment on Streamlit.
- Sentence transformer embeddings for semantic profile matching
- Cosine similarity ranking with configurable priority weights
- Live at hrmiitmaa.streamlit.app
References
- Prof. Ranjit Bauri
Associate Professor, Dept. of Metallurgical & Materials Engineering, IIT Madras. MS thesis supervisor. Available upon request.