Team & OnboardingAI

Team

• Nico Sharei
• Kiduk Kang
• Artem Bisliouk
• Marc Becker
• Ludwig Fügner

Solution

Thanks to the great teamwork I am proud to report that we were able to implement OnboardingAI in 24 hours.

OnboardingAI is a service that generates a personalized step-by-step onboarding process using LLMs together with Retrieval Augmented Generation based on company, employee and specific management requirements.

Project structure

• Presentation of OnboardingAI
• Frontend (Flutter web client)
• Backend(Flask. LLM + RAG)

Challenge Source

Original source: https://github.com/Q-hackathon/IBM_Q-Hack2024?tab=readme-ov-file

The Challenge

Artificial Intelligence for Evidence-Based Decision-Making in Public Spaces: Develop algorithms and models to analyze data from various sources to support informed decisions in policy areas such as healthcare, education, and environmental protection.

Introduction

Imagine you're at the forefront of the public sector, where the weight of your decisions can ripple through communities, shaping lives and our environment. Yet, amidst this vast responsibility, you're often navigating through a fog of complex data, bureaucratic processes, as well as urgent timelines and high-stakes scenarios. It's a delicate balance: acting swiftly without sacrificing accuracy or depth. Picture a future where AI assists you in quickly identifying patterns, predicting outcomes, highlighting the most impactful decisions. Bring the public sector into a new era of evidence-based decision-making with AI as your cornerstone.

Outcome & Deliverables

• A Proof of Concept showcasing the feasibility of the solution and technologies applied.
• A user-friendly interface tailored to a specific audience (Health, Environment, or Education)
• A presentation deck that covers:
  • The addressed issue and an overview of the proposed solution.
  • A diagram illustrating the solution's architecture.
  • A video recording showcasing a demo of the prototype.
  • An evaluation of the solution's effectiveness and relevance.
  • An outlook detailing how the solution could be scaled and further developed.

Challenge Requirements

What Technologies/Tools can be used?

• Ideation / Design / Documentation
  • Prototyping (Figma, InVision, XD …)
  • Diagrams (e.g., draw.io …)
• Implementation
  • Large Language Models, Multi-Modal-Models
  • Retrieval Augmented Generation, Embeddings, Semantic Search, Finetuning, GenAi + ML (hybrid approaches), Prompting
  • Python, Pandas, LangChain
  • ML Pipelines, MLOps