Pravar Agarwal

Engineered and deployed 15+ robust Python and bash automation scripts on linux systems to eliminate repetitive backend operational tasks, reducing manual checks by 25% and significantly improving process reliability. Utilized Python libraries, like Pandas, to automate critical data processing, cleaning, and transformation workflows for 1000+ daily financial CSV files, ensuring file generation integrity across various regional trading operations. Monitored and maintained real-time data flow supporting 3 major regional trading operations, ensuring the timely and secure generation of 100+ critical financial files daily.

Scaled the "Special Notice" reporting application by architecting and implementing regional support for Salt Lake using vanilla Java, enabling critical audience measurement reporting for Tier-1 clients such as Netflix and Hotstar. Decreased production defects by 15% through comprehensive refactoring and Object-Oriented (OO) design; leveraged Informatica to gain deep visibility into ETL pipelines, significantly accelerating the debugging of regional data flows.

Worked for a newspaper company and created various AI Agents using frameworks like Groq and CrewAI to implement multiple use cases which, in turn, increased content generation by 15% and reduced article production time from 25 min/article to 15 min/article. Vertex AI: Created Google Cloud functions and accessed Google’s foundation models to work on various use cases. Increased the traffic for the company’s English Newspaper by roughly 150%.

Designed and implemented NLP workflows leveraging OpenAI models, cutting data processing time by 40% and improving real-time insights delivery.