Software Engineer
Current- Part of the Generative AI Platform team.
- Formed a versatile RAG (Retrieval Augmented Generation) pattern migrating from text based search to semantic hybrid search.
- Used Azure Cognitive Search and OpenAI's GPT-3.5 and GPT-4 models to perform semantic hybrid search on text based and vector based indexes.
- Created an endpoint that specifically converted standalone question to VectorEmbeddings using EmbedQuery functionality from LangChain.
- Consumed the vector embedding that came as a response from the API built to perform Hybrid Search which improved the response time by 30% as compared to text based search.
- Made GPT model as a customizable property for end users to have a choice of the model from which they want answers.