At DotPe Private Limited, I built DotPool and DemandPulse, demand forecasting and decision intelligence platforms for restaurant merchants. DotPool used LightGBM to forecast daily orders and average order value across restaurant formats and Indian cities, achieving an R² of 0.935 on a time-split holdout.
I also developed a Python integration syncing Zoho CRM with Google Sheets every 30 minutes and deployed it with GitLab CI/CD, Jenkins, and Kubernetes. My work includes ETL/ELT pipelines processing 100K+ records daily, REST APIs, and automation using Slack bots and Gmail.
Across my projects, I’ve integrated LLM assistants with forecasting tools and built an agricultural diagnostic AI pipeline at E&ICT, IIT Kanpur. That project combined YOLO-based image detection with a Llama 3 retrieval pipeline to reduce diagnostic report generation from manual hours to under 30 seconds.
