Viraj Shah

Data Scientist @ CyberShrey LLC | MS in Computer Science | Machine Learning

Data Scientist | Cybershrey LLC Jul 2024 – Present

• Developed and deployed a real-time demand forecasting model using LSTM networks to optimize liquidity management and operational efficiency across global markets, improving prediction accuracy by 35% and reducing operational costs by 20%
• Architected a fraud detection system leveraging Graph Neural Networks (GNNs) to detect anomalies and fraudulent activities in large-scale financial transactions, achieving a 96% recall rate, and significantly reducing financial risk
• Implemented a personalized recommendation engine using collaborative filtering and deep learning to enhance customer experience and drive engagement, resulting in a 22% increase in user interaction with financial products and services
• Designed and optimized a scalable data lake architecture on AWS (S3, Glue, Athena) to centralize and streamline data storage and processing, reducing data retrieval latency by 40% and improving accessibility for global analytics teams
• Automated feature engineering pipelines for time-series and structured data using Python and Apache Spark, accelerating machine learning model development by 25%, enabling faster deployment of predictive analytics solutions

• Engineered a real-time anomaly detection system using Autoencoders and Isolation Forest to improve enterprise server reliability, reducing critical incident response times by 45%
• Optimized and deployed deep learning models on Azure ML and TensorFlow Serving, achieving a 35% improvement in inference speed and automating business intelligence workflows to cut decision-making time by 30%
• Developed an NLP pipeline for email classification and auto-tagging, reducing client response times by 30%, and built an internal recommendation system to enhance cross-department collaboration
• Implemented model monitoring and drift detection frameworks to ensure continuous ML performance and contributed to AI ethics protocols, promoting fairness, accountability, and transparency in enterprise solutions

• Developed a customer lifetime value (CLV) prediction model using Gradient Boosting, increasing client retention by 12% and improving marketing ROI through more effective outreach strategies
• Conducted customer segmentation analysis with K-Means and Gaussian Mixture Models, enhancing targeted marketing efforts and driving a 15% increase in conversion rates for property investments
• Built and deployed an NLP-based document classification system using BERT, achieving an F1 score of 0.93, significantly improving internal document retrieval and accelerating deal evaluations
• Implemented automated data pipelines with Apache Airflow to process and analyze 50K+ client records, saving 15 hours per week and enabling faster, data-driven decisions in property acquisitions and loan processing

MS in Computer and Information Science, New York Institute of Technology | GPA: 3.73 Sep 2021 – May 2023

B Tech in Electronics and Computer Engineering, Vellore Institute of Technology| GPA: 3.75 Jul 2017 – Jun 2021