Featured Project

UGC-AI — AI SaaS for Ad Creative GenerationAn AI-powered SaaS that generates UGC-style ad creatives from a product image and a person — transforming static inputs into lifestyle visuals and short-form marketing videos.

View RepositoryLive Preview
Active Development

Tech Stack

Next.jsReactTypeScriptNode.jsExpressMongoDBPrismaCloudinaryClerkOpenRouterReplicateGoogle Gemini AIAxiosTailwind CSSSentry

Architectural Decisions

Designed a multi-step AI pipeline (image → video) instead of isolated generation, enabling real marketing workflowsDecision 1 • Key architectural choice
Integrated multiple model providers (Gemini, OpenRouter, Replicate) to avoid vendor lock-in and optimize cost vs qualityDecision 2 • Key architectural choice
Switched from Gemini to free-tier models due to pricing constraints, maintaining product continuity under budget limitsDecision 3 • Key architectural choice
Handled asynchronous processing for long-running AI tasks with retry logic and failure recoveryDecision 4 • Key architectural choice
Implemented a credit-based system aligned with compute-heavy AI operationsDecision 5 • Key architectural choice
Designed the system to be model-agnostic, allowing future upgrades without architectural changesDecision 6 • Key architectural choice
Managed inconsistent AI outputs with validation and fallback handling to prevent UI breakageDecision 7 • Key architectural choice
Deployed backend on Render, accounting for cold-start latency in UXDecision 8 • Key architectural choice
UGC-AI • Live Preview
UGC-AI — AI SaaS for Ad Creative Generation Preview

Project Timeline

Status
Active
TypePortfolio System
ComplexityAdvanced
ArchitectureFull-Stack

Development Approach

Built with server-first architecture using Next.js App Router

TypeScript end-to-end for type safety across all layers

Centralized validation with shared Zod schemas

Explore the Codebase

Dive into the implementation details, architecture patterns, and development decisions.

GitHub Repository →