Data Scientist and AI/ML Engineer with experience building end-to-end ML pipelines, multimodal AI systems, ETL data infrastructure, and production dashboards across large-scale datasets. Proficient in Python, SQL, PyTorch, TensorFlow, and cloud platforms (AWS, GCP, Azure) for data engineering, model development, MLOps, LLM fine-tuning, computer vision, NLP, and BI reporting. Published researcher and cross-functional Agile collaborator. MS Data Science, GPA 3.7/4.0.
Xyton Semiconductor
Fine-tuned Vision Transformer (ViT), YOLO, and ResNet models on domain-specific circuit diagram datasets using Roboflow and PyTorch, improving component classification accuracy by 18% for Analytical Circuit Design applications.
Architected scalable AWS ETL pipelines (S3, Lambda, Glue) for LLM pre-training and fine-tuning across multi-million-record datasets, reducing data processing time by 40%.
Built end-to-end netlist generation pipeline integrating YOLO, ResNet, and Transformers to fully automate circuit diagram-to-SPICE conversion, replacing a previously manual EDA workflow.
Performed data annotation and tagging of 500+ circuit diagram images using Roboflow to create high-quality training datasets for computer vision model development.
Automated IEEE paper metadata extraction using web scraping and REST API integration, processing 500+ research documents to support a multimodal LLM knowledge platform.
BMW Manufacturing Co., LLC
Developed complex production routing dashboard using Oracle APEX, SQL, PL/SQL, and Oracle Database tracking 20+ workflow KPIs, reducing manual reporting time by 70% and eliminating routing errors on the Band F1B assembly line.
Enhanced Manager’s Overview dashboard with dynamic color-coded KPI tracking and conditional formatting logic displaying real-time shift-target performance (under-target, at-target, over-processing) across H50 and H52 production facilities.
Optimized PL/SQL queries for production and defect dataset extraction, enabling real-time, accurate reporting for cross-functional engineering teams and executive leadership.
Conducted hands-on process observations on Band F1B assembly line, identified workflow control gaps, and implemented corrective vehicle routing measures to eliminate production delays.
Indus University
Designed end-to-end ML classification pipeline on 100,000+ patient records using Python and Scikit-learn, engineering 40+ clinical features and reducing model inference time by 45% through hyperparameter tuning.
Achieved 92% accuracy in heart failure prediction using Random Forest, XGBoost, and SVM; published results in a Web of Science-indexed journal (Feb 2024) and presented at a national academic conference.
Conducted comprehensive EDA and statistical analysis on clinical datasets, applying hypothesis testing, regression analysis, and feature selection to improve model robustness and interpretability.
Built interactive Tableau dashboards visualizing clinical prediction outcomes and model performance metrics for medical researchers and senior stakeholders.
Mentored junior team members on ML methodologies, model evaluation frameworks, and research documentation standards.
Towa Pharma Pvt. Ltd.
Automated HR analytics workflows for 300+ employee records using Python and SQL, improving analytical processing speed by 75% across 20+ departments.
Built Power BI dashboards for real-time workforce trend monitoring, contributing to a 5% reduction in voluntary attrition and $200K+ in annual savings through data-driven HR decision-making.
Developed ML classification models and automated data pipelines to identify employee attrition risk factors, supporting cross-functional HR strategy and executive reporting.
Delivered analytical insights to HR leadership and senior management to guide data-driven workforce planning, policy changes, and talent retention initiatives.
University at Albany (SUNY Albany)
Indus University, India
Developed ML pipeline integrating BLIP and Falcon models for image understanding and text generation. Built full-stack web application using Anvil framework with custom API endpoints. Implemented robust error handling and optimization for production deployment.
Engineered ML system with 89% prediction accuracy using ensemble learning (Random Forest + XGBoost). Reduced model inference time by 45% through optimized feature engineering pipeline. Processed 100,000+ song records with 40+ features for comprehensive analysis.
Built end-to-end ML pipeline for employee attrition prediction using Python and scikit-learn. Performed advanced feature engineering and selection to identify key attrition indicators. Developed interactive dashboards using Jupyter and Plotly for real-time HR analytics.
Created ML models achieving 90%+ accuracy in early detection of cardiovascular diseases. Implemented multiple classifiers (Random Forest, XGBoost, SVM) for optimal prediction. Developed interactive dashboards for real-time model performance monitoring.
Indian Journal of Natural Sciences (Web of Science Scopus Journal)
International Journal of Research and Analytical Reviews (Impact Factor: 7.17)