Learn By Doing - Authoring Tool

LBD Authoring Platform

Executive Summary

As a solo developer during my internship at USDC Global, I led the end-to-end design and deployment of a full-stack authoring platform that revolutionized the company’s content production. The platform, built with Next.js and FastAPI, transformed a 5-hour, multi-stakeholder workflow into a 20-minute, self-service task for a single person. This represents a 93% reduction in total human effort, empowering 40+ Subject Matter Experts (SMEs) to create and deploy dynamic learning modules that now impact over 2,000 students.

The Problem: A Resource-Intensive Content Workflow

USDC Global’s patented “Learn by Doing” (LBD) modules are a core pedagogical asset. However, the creation process was a significant drain on resources and a bottleneck to scaling the curriculum. The legacy workflow was a complex, multi-step process requiring a minimum of 5 hours of combined effort from key, high-value personnel for every single module:

SME Specification: An SME would ideate a module and spend time writing a detailed technical specification document.
Developer Implementation: A software developer would interpret this document and custom-code the entire interactive module in React. This step was prone to misinterpretation of complex academic logic.
Review and Revision Cycle: The developer’s implementation was sent back to the SME for approval. This often initiated a lengthy back-and-forth cycle of feedback and revisions.
Manual Deployment: Once finally approved, the module had to be manually pushed and integrated into the company’s University Management System (UMS).

This entire process was slow, expensive, and created a critical dependency on the engineering team, preventing the company from producing content at the desired scale.

The Solution: A Self-Service Content Engine

To solve this, I designed and built an internal, web-based authoring platform that abstracts away all technical complexity. The platform empowers the true experts—the educators and SMEs—to build, test, and deploy sophisticated learning modules themselves, entirely eliminating the need for developer involvement in the creation process.

The platform provides an intuitive, form-based interface where an SME can translate their educational vision into a fully functional application. Key capabilities include:

Dynamic Variable Creation: SMEs can define variables and set rules for their generation, such as a random integer within a range or a selection from a predefined list.
Versatile Question Authoring: The tool supports a wide array of quantifiable question types, from complex mathematical formulas and dependent calculations to interactive UI elements like multiple-choice, drag-and-drop, and string matching.
Live, Real-Time Preview: A core feature of the platform is a 1:1 live preview panel. As the SME defines variables and builds tasks, the final interactive module is rendered instantly. This “what you see is what you get” approach removes the entire review/revision cycle.
AI-Assisted Content: To further streamline the process, I integrated AWS Bedrock, allowing SMEs to automatically generate explanatory text for solutions based on the variables and logic they’ve already defined.

The authoring tool directly generates a standardized configuration for the LBD, making the SME the sole creator and approver in one seamless workflow.

Technical Architecture & Key Features

I architected a decoupled, full-stack system designed for scalability, reliability, and ease of use. The entire system was built and deployed on AWS.

Frontend: A highly interactive and intuitive UI built with Next.js (React) and Material UI.
Backend: A robust FastAPI (Python) server that handles business logic, validation, and generation of the standardized JSON configuration.
Automated Build & Deployment Pipeline: I engineered a custom, event-driven build pipeline using Python scripts and containerized environments. When a module is finalized, the JSON is saved to an S3 bucket, which triggers a build process that compiles the module into a standalone HTML/CSS/JS bundle and deploys it to a separate S3 bucket.
Cloud Infrastructure: Leveraged AWS S3 for scalable storage of both JSON configurations and final deployable bundles.
Assessment and Analytics: Each generated LBD is pre-instrumented with hooks for logging student submissions and scores.

My Role & Key Challenges

I majorly this project from conception to launch. My collaboration with my mentor, Dr. Raja Subramanian, during initial brainstorming and constructive feedback during the technical architecture, implementation, and feature-scoping process helped immensely.

Leadership & Communication: I presented and demoed the platform prototype to the VP of Learning Innovation, successfully articulating its business value and securing executive buy-in for its adoption across the company’s content teams.
System Design & Abstraction: The most significant technical challenge was analyzing dozens of existing LBDs to design a flexible JSON meta-framework capable of supporting all academic disciplines.
User-Centric Iteration: I engaged my teammates as alpha testers, running usability tests to gather feedback and rapidly refine the application before its official rollout to SMEs.

Impact & Results

The LBD Authoring Platform has fundamentally transformed the content creation pipeline at USDC Global.

Drastic Workflow Efficiency: Replaced a 5-hour, multi-stakeholder process with a 20-minute, single-person workflow, representing a 93% reduction in total human effort per module.
Successful Company-Wide Adoption: Successfully onboarded 40+ Subject Matter Experts who now create and manage content autonomously.
Increased Content Velocity: Enabled the creation of 70+ new, standardized LBD modules within the first few months post-launch.
Broad Student Impact: The platform’s output directly impacts the learning experience of over 2,000 students across multiple degree programs.
Strategic Resource Allocation: Eliminated the engineering team as a bottleneck, freeing developer resources for core platform initiatives.

Impact Points/Summary

Engineered a full-stack Authoring Platform (Next.js, FastAPI) that transformed a 5-hour, multi-stakeholder workflow into a 20-minute self-service task, cutting total effort per module by 93%.
Architected and implemented a custom CI/CD pipeline using Python scripts and AWS S3, automating the build and deployment of interactive learning modules from a JSON configuration to serve over 2,000 students.
Led the project from concept to launch, single-handedly developing the application and successfully pitching the final product to the VP of Learning Innovation, securing its adoption as the company-wide standard.