NeMo Framework + Firebase Firestore and GCP - A newbie journey

Hello NVIDIA Developer Community,

I am both encouraged and delighted to be a part of the NVIDIA Inception program, and I am looking forward to building many innovative solutions together.

We are currently developing an application using FlutterFlow with Firebase as our backend, leveraging Firestore for real-time data storage. The core workflow involves capturing user data through the app and storing it on Firestore.

Our next endeavor is to process this data using the NVIDIA NeMo Framework to harness its powerful natural language processing capabilities. Additionally, we are implementing NeMo’s guardrails technology to enhance the accuracy of our services. We already have set up initial flows, intents and actions on colang language, and we are planning to utilize pre-trained models.

Moreover, our system will need to synergize with other GCP services, taking advantage of NVIDIA’s AI prowess. We are reading a lot and trying to find a way to make this whole system works, and already know a little bit about the technologies:

  • NVIDIA Triton Inference Server: For serving optimized, production-ready AI models.
  • Google Kubernetes Engine (GKE): For container management and orchestration.
  • Google Cloud Run: For running stateless containers in a managed environment.
  • Google Virtual Machine Image (VMI): For customizing and deploying VMs on the GCP.

As we are in the MVP phase, we are not looking to train our own AI models just yet, but instead, we aim to leverage the pre-trained models provided by NeMo.

**

Based on this information, I am seeking suggestions on:

**

  1. What would be the most cost-efficient and scalable path forward for integrating these technologies?

  2. Are there any best practices or specific considerations to ensure a seamless integration between FlutterFlow, Firebase, and NVIDIA NeMo?

  3. Which GCP services would you recommend prioritizing for an MVP that aligns with the capabilities of the NeMo Framework? Fast deploy, not much costs, just for the first tests.

Any insights, shared experiences, or advice on best practices for this integration would be greatly appreciated. I am particularly interested in ensuring that our implementation is cost-effective, efficient, and well-positioned to scale with our project’s growth.

Thank you in advance for your time and assistance!!