cv

Research Intern (Remote)

Indian Institute of Technology, Kharagpur

• Supervisor: Dr. Ishita Maiti
• Developing a GNN-driven multi-class segmentation pipeline for lung CT and chest X-ray scans, leveraging graph-structured representations for pathological region delineation.
• Designing graph neural network architectures to model spatial dependencies between anatomical regions for robust cross-modal medical image segmentation.

AI Development Intern (On-site)

Farmience AgroTech

• Built a Deep Q-Network agent within a Federated Learning framework to optimise irrigation scheduling and nutrient delivery for precision agriculture.
• Developed a computer vision pipeline for real-time crop anomaly detection integrated into autonomous agricultural systems.

Student Researcher (Part-time)

DST-ANRF (SERB) Funded Project, SRMIST

• Supervisor: Dr. Varadaraju Magesh

  Grant No: CRG/2022/005834

• Delivered a validated clinical prototype for real-time orthodontic debonding force measurement using FSRs, custom PCB, and Arduino Mega 2560 — published in Trends in Biomaterials & Artificial Organs (2026).
• Bridged embedded sensing and clinical orthodontics for quantitative intraoral pain assessment.

Business Intelligence & AI Strategy Intern (On-site)

SRM Technologies Private Limited

• Conducted research and competitive analysis of pharmaceutical BPO companies based in Japan, evaluating market positioning and operational capabilities.
• Performed research on Agentic AI for tender management automation, identifying opportunities for document processing and workflow optimisation.

Research Intern (Remote)

National Institute of Technology, Kurukshetra

• Supervisor: Dr. Ghanapriya Singh
• Developed FedDQPrune, a DQN-driven adaptive pruning framework for federated soybean yield prediction — RMSE 3.57 bu/acre, R² 0.87, outperforming FedAvg and FedPruning baselines; extended into FedDQN-Ctrl (SiMPaCT 2026).
• Reduced model size by 28.7% (628K → 448K parameters) with 28.7% communication savings via mask-aware federated aggregation for edge agricultural IoT deployment.

Research Intern (Remote)

Indian Institute of Technology, Jammu

• Supervisor: Dr. Badri Narayan Subudhi
• Developed a tri-modal forgery detection framework (ResNet-50 visual + FFT frequency + HOG gradient) with TriModalCross-Attention fusion — 69.56% test accuracy and AUC-ROC 0.7614 on the RTM dataset.
• Deployed a 3-head ensemble meta-classifier with adaptive architecture scaling and EMA weight averaging, outperforming Generic CNN and ManTra-Net baselines.

Industrial Trainee

Intel Corporation — Unnati Industrial Training Programme

• Built TRACE, an end-to-end IoT traceability pipeline using ESP32 + GM805L barcode scanner and YOLOv11n PCB defect detection across 7 defect classes (>95% precision, >85% mAP@0.5).
• Automated scan-to-result pipeline with barcode verification, defect classification, MongoDB audit logging, and a live Streamlit monitoring dashboard.

Research Intern (Remote)

National Institute of Technology, Tiruchirappalli

• Supervisor: Dr. G.R. Gangadharan
• Designed SMT-UPerNet for liver tumour segmentation on LiTS17 — 89.00% Dice, 80.20% IoU, outperforming Swin-Unet by +0.60% Dice. (HINT 2025, Scopus-indexed)
• Integrated Grad-CAM into the segmentation decoder with progressive layer-unfreezing and EMA training, achieving HD95 3.28 mm boundary accuracy.

Research Intern (Remote)

Indian Institute of Information Technology, Allahabad

• Supervisor: Dr. Ramesh Kumar Bhukya
• Fine-tuned AI4Bharat IndicConformer (129M params) for low-resource Kashmiri ASR using NeMo with Hybrid CTC/RNNT decoding, reducing WER from 39.1% to 14% over 50 epochs.
• Built dynamic audio chunking, regex-based Kashmiri text normalisation, and a custom WER callback for real-time training monitoring across a 5,632-token multilingual vocabulary.

B.Tech

SRM Institute of Science and Technology

Electronics and Communication Engineering (Data Science Specialization)

• Registration No: RA2311053010090
• Specialization in Data Science