Overview

• BioQuery AI Assistant is a specialized chatbot designed for clinical research organizations, with a focus on cancer drug discovery. It provides seamless access to diverse research data stored in multiple formats, including PDFs, CSVs, PowerPoints, and Excel sheets, for managing data. Powered by Retrieval-Augmented Generation (RAG) and MongoDB, BioQuery simplifies complex data querying and aggregation.
• Key features include automatic PowerPoint creation for research reporting and real-time inventory checks, enabling efficient workflows while ensuring strict data privacy compliance. Additionally, the tool offers real-time, word-by-word conversational capabilities similar to Alexa, enhancing user interaction and engagement.
• Data Integration: Merging data from varied formats into a unified system.
• NLP Accuracy: Handling complex, domain-specific queries with precision.
• Data Privacy: Ensuring secure handling of sensitive research data.

Key Features

• Accurate, reliable answers to experimental and research data queries.
• Automatic PowerPoint creation from raw input data.
• Real-time inventory checks powered by a language model.

Solution

• Precise Query Responses: Integrated RAG for parsing structured text data.
• Deployed MongoDB for efficient aggregation of non-textual datasets, ensuring accurate answers to complex questions in fields like biotechnology and biochemistry.
• Automated PowerPoint Creation: Leveraged python-pptx and Anthropics Claude 3.5 Sonnet to dynamically generate professional PowerPoint presentations from raw user data.
• Inventory Checks: Combined inventory databases with a language model to verify the availability of materials, enhancing lab efficiency and experiment readiness.

Technology Stack

• Programming Language: Python
• Qdrant: Vector database management
• MongoDB: Data aggregation
• FastAPI: Web interface development
• AWS Bedrock: Foundational ML models
• EC2: Scalable compute resources
• S3: Secure data storage
• EKS: Containerized application management
• Infrastructure Management: Terraform