Professional Experience
Cyberinfrastructure for Network Science Center, Indiana University
Machine Learning Research Assistant
Mentor: Dr. Katy Börner
September 2025 - Present
- Developed a 3D biomedical image-segmentation pipeline for the NIH HuBMAP consortium using SAM and DeepCell-Types, producing high-resolution cellular representations for downstream tissue-level analysis.
- Built automated data and model pipelines with MLflow and containerized them using Docker to ensure full reproducibility.
- Improved PyTorch imaging workflows through quantization and GPU-optimized inference, increasing model stability and throughput for high-volume multi-omics biomedical data.
Jio Institute, Mumbai
Data Scientist
Mentor: Dr. Sudipta Roy
July 2024 - June 2025
- Created a self-supervised learning pipeline using spectral features and attention for USG representation learning.
- Achieved 5–7% improvement in diagnostic accuracy across multi-organ datasets by enhancing feature generalization.
- Collaborated with clinicians at 3 government hospitals to curate and analyze over 5,000 patient cases, identifying dietary patterns related to fatty liver disease.
K. J. Somaiya College of Engineering, Mumbai
Research Intern
Mentor: Dr. Jagannath Nirmal
Aug 2023 - June 2024
- Developed a wavelet-based multi-resolution analysis framework for crown-of-thorns starfish (COTs) detection, achieving 95% classification accuracy, surpassing prior methods by 2.5%.
- Published findings in IMTS Springer Journal, showcasing advancements in integrating GLCM-based texture analysis with the second-order statistical feature extraction.
Ernst & Young (EY)
Software Development Intern
Manager: Ms. Nidhi Kumari
June 2023 - Aug 2023
- Developed the backend for a speech-to-text pipeline with speaker diarization on GCP, enabling agent scoring through voice-based profiling with 85%+ segmentation accuracy.
- Crafted a tone analysis module using speech embeddings and linguistic markers to detect misselling in 1,200+ calls.
- Enhanced RCM (Risk Control Matrix) using a fine-tuned RoBERTa model, reducing compliance review time by 30% via automated risk signal extraction.
Indian Statistical Institute, Kolkata
Research Intern (Biomedical Imaging and Bioinformatics Lab)
Mentor: Dr. Pradipta Maji
May 2022 - July 2022
- Developed advanced color normalization techniques for H&E stained images, utilizing RGB, HSI, and Lαβ models.
- Incorporated Gaussian Mixture Models to reduce noise and applied non-linear normalization to enhance image quality.
- Utilized the Akaike Information Criterion (AIC) to confirm HSI as the optimal color channel for enhancing breast cancer cell detection accuracy.