portfolio

Epoch AI

AI application installation made easy

Andrew

19 Jun 2025 — 2 min read

Overview

Client name: Benedict Lee
Budget: $3.000
Timeline: 3 month

About the project: Epoch AI is built as an AI solution app store, one of the first of its kind. Benedict got the idea while installing application on his mobile phone, and think to himself: what if adding AI into your application is as simple as clicking an Install button.

Value proposition: The application is designed with usability as the utmost important facet, gearing towards user without a lot of technical knowledge. Epoch AI aims to lower the barrier of entry to AI technology to users of all technical level.

Main challenges: AI is not an easy concept to simplify, so walking a tightrope between keeping it intuitive and still ensuring enough functionality present lots of challenge. The other problem is to develop an app that is efficient to run, by employing a Serverless architecture on AWS

Notable features

Epoch AI present several features as "App". By "installing" those app, you can set up an installation with configuration & integrate it into your own websites. Each app is a different AI system, with a different "installation" and "usage" process.

AI Agent

AI Agent is a RAG solution for question answering in a knowledge base. The installation process involves uploading documents into a knowledge base, then pick a LLM such as Claude Haiku, Mistral, Llama. The output is an embeddable Chat UI to put in your own website.

Insight AI

Insight is a quick & easy solution for data forecasting. User can upload a tabular dataset for the AI to analyze, then ask the AI for precise, data-backed prediction or overall trend analysis.

System Architecture

The requirements of running the app efficiently was solved by employing a fully serverless architecture on AWS. The upside to this approach is:

Pay what we use. Close to zero cost during idle period.
Near infinite horizontal scaling. Handle spike period with ease.
Global scale by leveraging AWS cloud, allowing low-latency independent of geographic obstacle

The architecture consists of:

React.js application served by Cloudfront from a S3 bucket.
Lambda functions behind API Gateway to handle application logic
DynamoDB as application data store
Cognito for managed authentication & seamless integration with other identity provider (Google, Meta login)

Containerizing Node.js application – Part 2

In the previous part of this series, we looked at how one may go about to improve the build time of your Node.js Docker container. In this part, I will go more in depth with regard to the size of your container. Why making your container lightweight A good

Containerize your Node.js application – Part 1

As one of the most popular backend tech stack at the moment, containerizing Node.js is still a chore for most developers. Here is my attempt to present a guide on how to do it, efficiently enough for production. The scope for Node.js based application is wide: * It can

My custom CV builder

I have a very customized CV building pipeline that helps me churn out a clean, customized CV in a matter of minutes. It was born out of necessity of when I went on job hunting and had to tailor each submission to the JD posted. I'm sharing because

How I manage my finance (as a developer)

We have all been there. You earn that lucrative IT-sector big buck, and next thing you know, you are down to 10% of your earning by the middle of the month. You don't even know how you spent the money. Then you can only vaguely recall buying some