AI/ML Intern

Overview

As an AI/ML Intern at DYU Labs (Ahmedabad, India), I served as technical lead for a summer intern team developing production-grade renewable energy intelligence systems. Our work was deployed for enterprise and government clients including ISRO, NTPC, and Aditya Birla Group.

Solar Forecasting & MLOps

Built an end-to-end forecasting pipeline for 75 production inverter devices across 15 solar plants, generating 15-minute interval day-ahead forecasts.

Key responsibilities:

• Telemetry ingestion from AWS DynamoDB, combined with weather forecasts from ECMWF and Copernicus ERA5 climate datasets
• Preprocessing for inverter telemetry normalization, outage-aware cleaning, and temporal/meteorological feature engineering
• Per-device LightGBM models with automated retraining, lifecycle management, and fleet-wide monitoring (MAE, MAPE, drift detection via Kolmogorov-Smirnov tests)
• Cloud deployment via AWS Lambda, EventBridge, ECR, and Fargate for fully automated pipeline execution

LangGraph-Powered Conversational AI Assistant

Led development of a LangGraph-powered assistant enabling solar plant operators to interact with telemetry systems using natural language and voice.

Key responsibilities:

• Multi-agent LangGraph workflows for query understanding, tool selection, validation, and response synthesis
• RAG pipelines using OpenAI embeddings, pgvector, and Supabase vector search
• NL-to-Database Query (NL2DBQ) for operational analytics and Modbus/register-based telemetry decoding
• Observability via LangSmith; secure tool execution with authorization guardrails