Summary of Ai Hackerspace Session - July 12, 2024

Reuven Cohen: Hi everyone, welcome to today's session of AI Hackerspace. I'm Reuven Cohen, and I'm excited to be here with all of you. Let's get started with today's agenda.

Reuven Cohen: 0:00 - We'll kick things off with a few presentations and then move on to some lightning talks and technical demonstrations. Feel free to share your social media links in the chat. This is a great opportunity for networking.

0:31:16 - Victor Gasnikov asked about deploying the Docker image of Hugging Face chat on Free tier. Rishub (Craftsman) recommended using Hugging Face for fine-tuning and deploying MVPs at an affordable price, mentioning alternatives like Replicate and RunPod for better inference speed and production readiness.

0:32:21 - Victor Gasnikov inquired about using someone else's API key on Hugging Face. It was clarified that using someone else's space would mean using their API key, and they would pay for the tokens.

Dimple Francis: 0:36:46 - Presented on enhancing AI chatbots with graph-based Retrieval-Augmented Generation (RAG) and demonstrated WimbleChat. Dimple explained how Neo4j can be used for storing embeddings and creating nodes and edges to improve chatbot functionalities.

Greg DeCarlo: 0:59:59 - Presented on creating HTML, CSS, and JS code from a single prompt using his site, You Say AI Do. He discussed different CSS libraries for client-side code and the rapid iteration capabilities of his tool.

Lightning Talks:

Dimple Francis: 0:49:08 - Discussed the benefits of using Neo4j for complex relationship retrieval in chatbots and provided a detailed explanation of the technical setup and real-world applications of graph-based retrieval.

Greg DeCarlo: 1:38:47 - Introduced a GitHub repository implementing 58 prompting techniques, highlighting practical applications in various AI projects.

Technical Demonstrations:

Dimple Francis: 0:49:08 - Demonstrated how to use Neo4j for storing embeddings and creating nodes and edges to enhance chatbot functionalities.

Greg DeCarlo: 1:14:00 - Showcased his site for generating front-end code from prompts and discussed the versatility and user-friendliness of his tool.

Discussions and Q&A:

Deployment challenges on Hugging Face: 0:31:16 - Discussed solutions for deploying AI models on different platforms, emphasizing the use of private environment variables and free hosting options.

Use of API keys on Hugging Face: 0:32:21 - Clarified API key usage and environment variables for secure and efficient deployments.

DevOps pipeline failures and rollback issues: 0:09:10 - Detailed discussion on best practices for handling driver issues and rollbacks.

Special Topics:

Dimple Francis: 0:49:08 - Presented on graph-based RAG for chatbots, explaining how to use Neo4j for storing embeddings and enhancing chatbot functionalities.

Stephen Wise: 1:21:43 - Introduced a project for creating tutorial videos for non-developers, focusing on making AI tools accessible to beginners.

Community Engagement:

Networking Initiatives: Encouraged sharing of LinkedIn profiles to enhance networking.

• Rishub (Craftsman)
• Victor Gasnikov
• Abhishek A
• Philippe Bogaerts
• Gideon Marken
• Ivan Djordjevic
• Bradley Ross
• Matt Paige
• Wilver Melendez
• Reggie Chan
• Bence Csernak
• Anil
• Likhith V
• Stephen Wise
• Joseph Seager
• Monish

Technical Challenges:

• Discussed the deployment of Docker images and API key management on Hugging Face.
• Addressed issues with the Microsoft Teams environment and device management for consultants working with multiple companies.

Future Call Topics:

DSpy Framework: 1:30:48 - Proposed for future discussion with a deep dive into its applications and tutorials.

Creating Tutorial Videos for Non-Developers: 1:21:43 - Introduced by Stephen Wise, focusing on making AI tools accessible to beginners.

Closing Remarks:

Summary of Discussions: Recapped the main presentations and discussions on AI tools and their applications. Highlighted key insights from the technical demonstrations and Q&A sessions.

Expressions of Gratitude: Acknowledgments were given to all participants for their contributions and engagement.

Future Plans: Mentioned plans for discussing the DSpy framework in future sessions and continuing the development of educational resources for non-developers.