Bachelor of Technology Student
Computer Science and Artificial Intelligence student with experience in machine learning, deep learning, computer vision, NLP, quantum computing, software engineering, and data-driven research.
Defence Research and Development Organisation (DRDO) – SAG
Jun 2026 – Jul 2026
Implemented three-qubit bit-flip and phase-flip QEC codes with lookup-table decoding, and distance-3/distance-5 rotated surface codes with MWPM decoding using Qiskit, Stim, and PyMatching. Achieved 3.7× logical error suppression at p=0.10 and localized the surface-code threshold crossing between p=0.5% and p=0.7%. Delivered a 12-page technical report and presentation with a roadmap toward full fault-tolerant resource estimation.
IGDTUW – Anveshan Foundation
Jun 2025 – Jul 2025
Trained a pneumonia classification model on the RSNA Chest X-ray dataset by fine-tuning a PyTorch MLP head on frozen TorchXRayVision DenseNet121 features. Addressed class imbalance with balanced focal loss, tuned hyperparameters and decision thresholds against validation F1, and integrated the trained model with the team’s Explainable AI module.
Taarangana, IGDTUW
2025 – Present
Associate in event management for the annual cultural fest.
Rotaract Club, IGDTUW
2024 – Present
Member involved in community services.
IEEE IGDTUW
2024 – 2025
Junior Executive responsible for sponsorship activities.
Bachelor of Technology, Computer Science and Artificial Intelligence
2024 – 2028
CGPA: 8.87; SGPA: 9.45
Class XII (CBSE)
2022 – 2024
94.4%
Class X (CBSE)
2020 – 2022
94.8%
Developed and trained a pneumonia classification pipeline on the RSNA dataset using a pretrained TorchXRayVision DenseNet121 feature extractor and custom PyTorch MLP head. Addressed class imbalance with focal loss, achieving 0.873 AUC-ROC and 0.643 F1-score, and integrated Grad-CAM, Grad-CAM++, and Integrated Gradients for interpretable predictions.
Built and benchmarked TF-IDF with Logistic Regression, LSTM, and fine-tuned DistilBERT approaches on 17,880 Kaggle job postings. Achieved 98.43% accuracy and 84.00% fraud-class F1 with DistilBERT, then deployed a quantized ONNX model client-side through a Hugging Face static Space using transformers.js/WebAssembly.
Built a Python research pipeline using per-ticker XGBoost return forecasts across a 47-stock NSE universe and a Black-Litterman posterior. Used purged walk-forward backtesting, diagnosed look-ahead bias with a random-walk control, and implemented regression testing, bootstrap confidence intervals, rank IC, and transaction-cost modelling.
Built the AI/NLP engine for a five-member coastal-emergency detection platform using zero-shot classification with BART and RoBERTa across 10+ hazard categories. Achieved 90%+ hazard-detection accuracy at 1,000+ posts per minute with multilingual Hindi, Tamil, Telugu, and English support.
Modeled Delhi’s road network as a weighted directed graph in Java and implemented Dijkstra, A*, and Yen’s K-Shortest Loopless Paths through a Spring Boot REST API and standalone CLI. Added road-closure toggling, traffic-multiplier edge weighting, and top-K alternative route generation.
Designed a Java workflow engine using the State Design Pattern and core OOP principles for multi-stage document approval pipelines. Implemented dedicated state classes, file-based persistence, and timestamped audit logging for traceable workflow transitions.
Built an interactive Python, Streamlit, and Plotly simulator implementing FCFS, SJF, SRTF, Round Robin, and Priority Scheduling. Computed per-process metrics, benchmarked algorithms, identified best performers, and created interactive Gantt-chart visualizations.
Designed a normalized MySQL database and synthetic dataset of 200,000 listening events across 5,000 users, 4,000 songs, and 150 artists. Wrote analytical SQL queries using CTEs and window functions and built a Python, Pandas, Matplotlib, and Streamlit analytics dashboard with Pytest and GitHub Actions CI.