Harshana Yadav

I built a Python RAG pipeline that uses document chunking, embeddings, and similarity-based retrieval to deliver context-aware answers. I evaluated response quality by tracking retrieval accuracy, groundedness, and unsupported-answer rates.

For Object Detection Using Gemini 2.5 Flash, I built a multimodal image-analysis workflow to identify objects and generate contextual descriptions. I assessed model outputs for identification, relevance, and consistency.

I also developed an AI Travel Assistant for route planning and alternative-route recommendations based on travel requirements. I tested travel scenarios to assess recommendation relevance and application behavior.

In a data analytics role, I built and optimized ETL pipelines with Python, SQL, and PySpark, and processed data using Azure Databricks. I also cleaned, transformed, and analyzed more than 10,000 records as a Data Analytics Trainee for the Government of Maharashtra.