Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


New on LowEndTalk? Please Register and read our Community Rules.

All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

Built a Gamified, Intent-Driven, AI-Assisted E-commerce on a 1GB VM — Would Appreciate Your Feedback

j1rd3ryj1rd3ry Member

Hey everyone,



Thank you for accepting me here. This is actually my first post on LowEndTalk, so I thought I'd share a project I recently finished and get some feedback from people with experience running applications on low-resource servers.

I recently built an e-commerce MVP for a small bakery client.

The project is a gamified, intent-driven, AI-assisted e-commerce storefront with a custom CMS.

The interesting part for me was seeing how much functionality I could run while keeping the infrastructure relatively small.

Current setup:

  • GCP e2-micro — 1GB RAM
  • Next.js storefront
  • Django / Django Ninja backend
  • FastAPI-based AI/intent layer
  • SQLite
  • Meilisearch
  • Nginx
  • Custom vanilla-JS CMS

For the MVP, I deliberately avoided adding infrastructure unless I felt there was an actual requirement for it.

No Kubernetes, no microservices everywhere, no Redis, no Celery, and no CDN.


The AI assistant

The storefront has an assistant called Tindera.

I didn't want it to be just another chatbot where everything is sent to an LLM and we hope for the best.

The basic idea is:

User request → intent recognition → application action

For example, the assistant can understand what the customer is trying to do and route the request toward things such as product search, navigation, or other storefront actions.

The LLM is only one component of the system. I don't fully trust an LLM to make application decisions on its own, so the application controls what actions are actually available.


Why Next.js?

This will probably be one of the questions. 😄

Yes, I could have built a lot of the frontend with vanilla JS.

The main reason I chose Next.js was SSR. When I considered the effort involved in building and maintaining the SSR/frontend infrastructure myself versus the additional framework/runtime overhead, Next.js seemed like a reasonable tradeoff for this project.

The 1GB VM is mainly an MVP/testing environment. The intended production environment would be around 4–8GB RAM, so I'm not trying to claim that 1GB is the ideal production size for this application.


What I'm interested in hearing from you guys

I wanted to see how far I could push a small VM before adding infrastructure simply because it's commonly recommended.

It actually worked better than I expected.

So I'd really appreciate feedback from people here who have experience running real workloads on low-resource VPSes.

I'm more interested in practical experience than theoretical architecture.

The goal wasn't to use the fewest technologies possible. It was more about finding a reasonable balance between development effort, runtime overhead, maintenance, and actual workload.

Anyway, I'm new here, so I'd genuinely appreciate any criticism or suggestions. 😄


Demo: http://34.68.53.234/

Demo credentials: demo / demo

Project / showcase: https://github.com/j1rDery/aireenshop-showcase

Comments

Sign In or Register to comment.