At Tata Capital Limited, I engineered a Python-based analysis engine for six credit scorecards, with JSON parsing, validation pipelines, and automated report generation. It cut manual processing effort by 90%.
I also built Selenium WebDriver workflows integrated with FinnOneNeo to automate Business Partner creation and approval for 500+ records, reducing manual effort by 90% or more. I designed rule-based branch mapping and product-allocation logic that cut QA effort by 80%.
In my WinOrbit project, I built a Windows-based Docker management system that responds to screen and system events by pausing or resuming configured containers. Runtime testing recorded 35–50 ms pause/resume latency, and I exposed monitoring metrics through Prometheus and Grafana.
I also built RailQueryAI, a RAG-based railway information backend that searches more than 100 policy documents with LangChain and FAISS. My work includes separate components for conversational queries, complaint classification with sentiment detection, and delay reporting.
