Deepak Mittal

- Architected and deployed a QA bot leveraging 12,000+ product PDFs, enhancing user inquiry resolution efficiency by 40%: Databricks, GPT, OpenAI, Azure, NLP - Engineered and optimized scalable backend pipelines, automating Information Retrieval and Embedding generation, reducing processing time by 50%: FastAPI, Docker, VectorDB, LLM - Led cross-functional teams to design and implement an Agentic RAG system with Google Vertex AI Agent Builder, reducing query processing time by 30% and enhancing system scalability by 50%. - Architected CICD pipeline, Docker container, and Artifact Registry pipeline for smooth Application Deployment on Google Cloud Run.

- Led the development of Intelligent Document Processing Studio(No-code model training and deployment product). Integrated more than 10 state of the art models for Document Classification & Extraction. - Developed backend pipelines to integrate various cloud-based data sources and down-stream data processing services. - Designed and developed the microservice-based architecture for data preparation, exploration, model training and model serving. - Leveraged Python, Java, and Linux to implement robust and high-performance solutions. - Experienced developing and deploying live production systems, as part of a product team. - Reduced processing time of a Java-based client component from around 120 seconds to less than 30 seconds.