Employee Onboarding

• 

  onboarding employees

Inspiration- The challenge of making employee onboarding more efficient, engaging, and personalized. Traditional onboarding can be overwhelming, with new hires drowning in documents. This tool aims to provide a structured, interactive, and less stressful experience by leveraging the power of AI.✨

What it does- This application transforms any URL containing onboarding information (like a Notion doc, Confluence page, or company handbook) into a structured, step-by-step onboarding plan. It provides a clear, digestible guide for the new employee and will allow them to ask questions about the material through an interactive chat interface, creating a personalized onboarding journey. 📝

How we built it- This project is built with a modern tech stack, leveraging Large Language Models (LLMs) for content processing and generation. Backend: A Python Flask server orchestrates the AI logic. It uses LangChain to structure interactions with AWS Bedrock (for embeddings and chat), Tavily for web content extraction, and MongoDB Atlas for vector storage and session management. Frontend: A responsive React application built with Vite provides a clean and intuitive user interface for submitting URLs and interacting with the generated onboarding plan. AI Core: AWS Bedrock's Titan and Claude models are used to create semantic embeddings of the source material and generate human-like, context-aware onboarding guides. Challenges we ran into- Integrating multiple cloud services (AWS, MongoDB, Tavily) and ensuring secure credential management. Debugging AWS authentication and model access permissions (InvalidClientTokenId, AccessDeniedException), which highlighted the importance of proper IAM configuration. Ensuring the generated onboarding plan was not just a summary, but a genuinely useful, step-by-step guide, which required careful prompt engineering.💻

Accomplishments that we're proud of- Successfully building a full-stack RAG (Retrieval-Augmented Generation) application from scratch. Creating a seamless workflow where a user can input a URL and receive a detailed, AI-generated onboarding plan in markdown format. Implementing a robust backend that correctly chunks, embeds, and stores document content for efficient similarity search.✨

What we learned- The power of prompt engineering in shaping the output of LLMs to create structured and high-quality content. Practical experience with the LangChain ecosystem for building complex AI workflows. The nuances of working with cloud-based AI services and vector databases, including authentication, permissions, and data modeling.💅

What's next for New Employee Onboarding- Implementing full conversational memory for the chat feature, allowing for more natural, context-aware follow-up questions. Adding support for more document types, such as PDF uploads. Creating a dashboard for managers to track the new hire's progress through the onboarding plan.

Built With

• amazon-web-services
• mongodb
• python
• temporal