My CV

Education

Research Projects

1. 3D High Dynamic Range Reconstruction of Dynamic Images using Gaussian Splatting

Jun’24-May’25
(M.Tech / Guide: Prof. Kaushik Mitra) IIT Madras
Keywords: 3D HDR Reconstruction, Point Clouds, Gaussian Splatting, NeRF, HDR

• Investigated the limitations of NeRF for HDR scene reconstruction and explored alternative representations for static and dynamic 3D scenes. Developed a comprehensive understanding of radiance fields, tone mapping, and multi-exposure fusion techniques.
• Proposed a novel dual-path pipeline combining Mertens exposure stacking with 3D Gaussian Splatting for real-time HDR rendering. Achieved PSNR of 37.129 on static scenes, outperforming HDR-GS (35.47) with superior fidelity and efficiency.
• Extended the 4D Gaussian Splatting framework to support 16-bit dynamic HDR inputs using a self-captured dataset. Enabled temporally consistent HDR synthesis for dynamic scenes. Also developed novel pipeline for HDR-Dynamic novel view synthesis

2. Analysis and Implementation of Machine Learning Model on CERN’s Big Datasets

Jun’22-May’23
(M.Sc / Guide: Prof. Seema Bahinipati) IIT Bhubaneswar
Keywords: Big Data Analysis, EDA, Machine Learning

• Handled petabyte-scale datasets from CERN, performed Exploratory Data Analysis to extract meaningful insights.
• Identified key features; optimized model input and enhanced the performance of the Boosted Decision Tree algorithm.
• The applied Boosted Decision Tree model achieved a 0.98 ROC value, effectively classifying signal and noise.

3. The Exponentially Changing Periodic Probability Density Function (PDF)

Nov’18-Mar’19
(B.Sc) K.T.H.M. College
Keywords: Probability Distribution, Mathematical Statistics

• Developed the Sinex distribution, a novel PDF with multiple peaks, using Laplace Transforms and MGF functions.
• Derived the Sinex distribution’s mean and variance using the gamma function and Inverse Laplace Transforms.
• Analyzed the Sinex distribution, showing it approximates the Gamma Distribution under certain conditions.

Personal Projects

1. Attention-Guided Hybrid Retinex for Low-Light Image Enhancement

Keywords: Image Enhancement, Image Restoration, Vision

• Designed a deep learning model that enhances dark, low-light images via Retinex-based decomposition, illumination relighting, and fusion.
• Integrated attention-guided illumination refinement and multi-scale residual learning for adaptive brightness correction and texture, color preservation.
• Achieved 21+ PSNR and 0.8+ SSIM on LOL-v1 with a full PyTorch pipeline for training, inference, evaluation, and visual diagnostics.

2. GAN-Based Multi-Exposure Prediction for HDR Imaging

Keywords: Generative Model, Deep Learning, CNN, HDR

• Developed a GAN-based model to predict a full 5-level multi-exposure image stack from a single LDR input for HDR-related imaging applications.
• Designed a convolutional encoder–decoder generator to synthesize photorealistic exposure variations that simulate real-world camera bracketing.
• Integrated a PatchGAN-style discriminator to assess global structure and local texture consistency across predicted exposure stacks.

3. Lung Cancer Detection Using CNN

Keywords: Classification, Medical Imaging, CNN

• Developed a CNN for lung cancer classification, achieving 92% validation accuracy on microscopic lung images.
• Achieved 97%, 87%, and 91% F1-scores for three lung cancer types, demonstrating strong model performance.
• Boosted efficiency with regularization & augmentation, achieving 92% macro-average precision, recall, & F1-score.

4. Autocorrect System

Keywords: NLP, Text Processing

• Engineered an autocorrect system that enhances text accuracy and user experience by suggesting correct words.
• Increased data processing efficiency by 15% by developing a scalable Python script for handling large text files.
• Devised a word variant production method to provide accurate spelling corrections by generating possible words.

5. Movie Recommendation System

Keywords: Data Preprocessing, Recommender Systems

• Built a content-based movie recommendation system using 4,800+ movies to provide personalized suggestions.
• Carried out data pre-processing and feature engineering on text and numerical data, thus boosting the accuracy.
• System retrieves and ranks the top 10 similar movies using a similarity matrix, enhancing the user experience.

6. Chatbot Creation

Keywords: Intent Classification, Text Embeddings

• Designed a Chatbot model using TensorFlow and Keras, achieving over 94% accuracy in intent classification.
• Mapped user input patterns to specific intents using a JSON-based dataset, enabling content-aware responses.
• Trained the model using embeddings, pooling, and dense layers, with the tokenizer and sequence padding.

Relevant Coursework

1. Pattern Recognition and Machine Learning (PRML) January 2024–April 2024

(CS5691) IIT Madras

• Built classification models like Naive Bayes, KNN, Logistic Regression, and Random Forest. Implemented Ridge, Lasso Regression, K-Means, PCA, DBSCAN, and GMM.

2. Computational Photography January 2024–April 2024

(EE5176) IIT Madras

• Worked on image reconstruction and enhancement techniques including demosaicing, denoising, alpha matting, motion deblurring, and compositing. Explored concepts such as Bayer pattern processing, tone mapping, HDR imaging, coded aperture, and light field imaging in computational photography.

3. Image Signal Processing (ISP) January 2024–April 2024

(EE5157) IIT Madras

• Performed image processing in Python: geometric transforms, mosaicing, shape from focus, invariant blurring, Otsu’s thresholding, and occlusion detection.

4. Natural Language Processing January 2024–April 2024

(CS6370) IIT Madras

• Built a Python-based search engine with NLP modules (segmentation, tokenization, lemmatization, stopword removal, spell correction) using NLTK, TF-IDF, and WordNet for semantic query matching.

5. Data Analysis & Visualization in R/Python/SQL (DAV) January 2024–April 2024

(MA5755) IIT Madras

• Built a neural network from scratch for handwritten digit recognition with 87% training and 86% test accuracy.

6. Probability Foundations for Electrical Engineers July 2023 – November 2023

(EE5110) IIT Madras

• Explored probability theory from fundamentals (expectation, variance, distributions) to advanced topics including joint/multivariate distributions, variable transformations, LLN, and probabilistic inequalities.

Online Coursework

• Data Structures and Algorithms in Python
• Computer Vision with OpenCV and Deep Learning
• Machine Learning, Data Science and Deep Learning
• Data-Driven Astronomy (University of Sydney)

Technical Skills

• Programming Languages: Python, SQL, C++, MATLAB
• Libraries: NumPy, Pandas, Matplotlib, TensorFlow
• Deep Learning & ML: Scikit-Learn, PyTorch, Keras
• Tools: MS Excel, PowerPoint, Word, Power BI
• NLP & Vision: NLTK, spaCy, OpenCV, Pillow, scikit-image, imageio

Positions of Responsibility

• Teaching Assistant: Designed quizzes and assignments for Modern Computer Vision course; supported 100+ students.
• Student Placement Coordinator: Coordinated placement season of IIT Bhubaneswar, handling over 200+ students.
• Assistant Coordinator, Counseling Service Team: Guided over 80+ freshers on-campus at IIT Bhubaneswar.

Achievements / Awards

• All India Rank - 617 in IIT-JAM 2021 Entrance Examination conducted by IISc Bangalore.
• First Prize in Poster Presentation Competition held at Sandip University, Nashik.
• Published 3 Scientific Articles during B.Sc.