Harshil Bhandari

Background

About Harshil Bhandari

I’m passionate about leveraging cloud computing, data engineering, and machine learning to transform complex challenges into scalable, real-world solutions. I’ve built robust ETL pipelines using Python—with Regex and Pandas on an Azure Databricks workspace—to process over 20 million rows of machine log data, and developed interactive Spotfire dashboards that turn raw data into actionable insights.I’ve spearheaded projects like the VeriShield Financial Verification Platform, where I designed a containerized microservices architecture using Docker, FastAPI, PostgreSQL, and Neo4j. I integrated Apache Kafka for asynchronous, event-driven workflows and enhanced fraud detection by developing a Graph Neural Network with PyTorch (leveraging CUDA) alongside models such as XGBoost and LightGBM. My deployment strategy included robust CI/CD pipelines on AWS—utilizing ECS, ECR, and Secrets Manager—to ensure scalable and secure operations.I also created the Polaris Multi-Agent Code Assistant, combining Flask microservices with AI APIs from OpenAI’s GPT-4o-mini and Google’s Gemini-1.5-Flash, orchestrated through Redis and Docker. My technical toolkit spans full-stack development with React, Node.js, and the Monaco Editor, and I’m well-versed in libraries such as NumPy, Scikit-learn, Keras, TensorFlow, OpenCV, PySpark, and NLTK. Proficient in C/C++, Java, and SQL, with hands-on experience across cloud platforms including Azure and Snowflake, I thrive on building innovative, data-driven systems that deliver measurable impact.