# Sankar Sanjay Varma Thotakura > Machine Learning Engineer | Python & FastAPI | Real-Time ML Systems | Fraud Detection & Predictive Analytics | AWS/Azure Location: Santa Clara, California, United States Profile: https://flows.cv/sankar I’m a Software Engineer and ML Engineer with 3+ years of experience designing and deploying ML-driven, cloud-native applications across insurance, industrial, and financial domains. My work focuses on building scalable backend systems, real-time data pipelines, and intelligent models that deliver measurable impact—such as reducing fraud by 15%, improving risk scoring by 40%, and enhancing system performance by 60%. Skilled in Python, FastAPI, Flask, PyTorch, TensorFlow, and AWS/Azure, I bring end-to-end expertise across backend development, machine learning, and DevOps automation. I’ve led projects involving real-time event streaming, NLP using Hugging Face Transformers, and CI/CD deployments with Docker, Jenkins, and GitHub Actions. I’m passionate about turning data into actionable insights, optimizing system performance, and building intelligent, reliable software solutions. Always eager to collaborate, innovate, and solve complex challenges that drive business and technology forward. ## Work Experience ### Software Engineer @ Berkshire Hathaway Specialty Insurance Jan 2025 – Present | Santa Clara, California, United States •Designed and deployed a real-time insurance fraud detection system using PyTorch, XGBoost, and LightGBM, processing over 2 TB of daily claims and policy data, reducing fraudulent payouts by 15% within the first quarter. •Built scalable ETL pipelines with Spark, Pandas, and SQL, integrating structured and unstructured data sources, enabling automated feature engineering and cutting manual preprocessing time by 40%. •Developed NLP-based models using Hugging Face Transformers to extract actionable insights from claim notes and customer communications, improving fraud detection precision to 92%. •Implemented containerized microservices with Docker and FastAPI, deployed on AWS ECS with CI/CD pipelines via Jenkins and GitHub Actions, achieving <200ms latency for real-time risk scoring. •Orchestrated large-scale ML workflows with Kubernetes and MLflow, automating model training, versioning, and deployment, ensuring 99.9% uptime and streamlined retraining for evolving data patterns. •Enhanced policy risk scoring accuracy using advanced ML models, enabling underwriters to prioritize high-risk accounts and reducing policy lapses by 40%, directly supporting strategic business decisions. •Collaborated with actuaries and data teams to translate model outputs into actionable insights, boosting business decision turnaround time by 35% through dashboard integration using Power BI and AWS Lambda. ### Software Engineer @ OXmaint- Maintenance Management Jan 2024 – Jan 2025 | Sunnyvale, California, United States •Developed modular backend services using Python (FastAPI, Flask) and SQLAlchemy, automating maintenance task workflows and reducing unplanned service downtime by 40% across industrial asset management systems. •Built real-time data ingestion and transformation pipelines using Pandas, NumPy, and scikit-learn, enabling predictive maintenance models that achieved up to 92% fault detection accuracy for HVAC and IoT sensor data. •Designed and deployed RESTful APIs with Swagger, OAuth 2.0, and JWT authentication, ensuring secure technician access and seamless integration with mobile field-service platforms. •Optimized SQL queries and indexing in PostgreSQL, improving analytics dashboard response times by 60% and supporting real-time reporting on maintenance KPIs. •Implemented containerized deployments using Docker and Azure DevOps Pipelines, ensuring consistent build environments and reducing deployment rollbacks by 70%. •Integrated ML-driven predictive reports into the analytics dashboard, helping clients achieve 25% reduction in equipment maintenance costs through proactive scheduling. •Collaborated with data science and QA teams to design automated test workflows using PyTest and Postman, increasing coverage and reducing production defects by 30% ### Software Developer @ Hexaware Technologies Jan 2021 – Jan 2023 | India •Developed and deployed RESTful microservices using Python (Flask, FastAPI) to process financial transactions and reporting workflows, improving system response time by 35% in enterprise banking applications. •Automated ETL pipelines leveraging Pandas, NumPy, and SQLAlchemy to cleanse, validate, and transform multi-source financial data, reducing manual data preparation time by 70%. •Integrated real-time event streaming using Apache Kafka and Redis, enabling fraud detection and alerting with sub-second latency across high-volume transaction systems. •Optimized SQL and NoSQL queries in PostgreSQL and MongoDB, enhancing query performance by 45% for credit risk and reconciliation modules. •Implemented API authentication and access control using OAuth 2.0, JWT, and Flask-Security, ensuring regulatory compliance for 100K+ secure sessions. •Containerized and deployed applications on AWS EC2 and Azure App Services using Docker and Jenkins, improving release consistency and reducing rollback incidents by 60%. •Collaborated cross-functionally with product, data, and QA teams in Agile sprints, integrating PyTest and Postman for API validation, which decreased post-release defects by 30%. ## Education ### Bachelor of Technology in CSE SRKR Engineering College ### Master's degree in Computer Engineering Santa Clara University ## Contact & Social - LinkedIn: https://linkedin.com/in/sankar-sanjay-varma-thotakura --- Source: https://flows.cv/sankar JSON Resume: https://flows.cv/sankar/resume.json Last updated: 2026-04-11