Machine Learning Engineer
Current-> Designed and executed end-to-end document ingestion process, ingesting and storing terabytes of company documents in Azure Blob storage.-> Created custom ingestion pipeline functionality to exclude confidential documents from the ingestion process.-> Orchestrated data preprocessing and cleaning procedures to ensure the quality and consistency of ingested documents, enhancing downstream processing efficiency.-> Integrated GPT-4 Vision models to extract additional context from images and tables within documents, augmenting textual information for richer understanding.-> Utilized advanced vector embedding techniques to create embeddings for text, images, and tables, enhancing the representation of diverse data modalities.-> Contributed in development of storage and retrieval mechanisms using Milvus Vector DB to efficiently manage and query vector embeddings at scale, ensuring rapid access to relevant information.-> Implemented a Multi-Vector Retriever for Retrieval-Augmented Generation (RAG) on tables, text, and images using Langchain, enabling the chatbot to generate contextually relevant responses.-> Integrated functionality for displaying source files alongside chatbot responses, allowing users to verify the relevance and correctness of provided answers.-> Applied appropriate machine learning algorithms and conducted rigorous tests and experiments to refine model performance.-> Contributed to the development of a robust feedback loop, enabling the chatbot to iteratively learn from user interactions and adapt its responses accordingly.
