AboutServicesPortfolioBlogContactGet a Quote
Home/Portfolio/Digital Wardrobe AI Pipeline
AI / Machine Learning2025

Digital Wardrobe AI Pipeline

Team Lead — ML-powered fashion recommendation engine

Team Lead
role
4
models
Real-time
pipeline

Exhibit Magazine India Pvt. Ltd.

Led and engineered the complete ML pipeline for the Digital Wardrobe App as Team Lead — integrating image styling models, outfit recommendation engines, and background removal pipelines to create an AI-powered fashion platform. Managed the team end-to-end from architecture through deployment.

PythonFastAPIML ModelsRunPodDockerREST APIs
The Challenge

Problem Statement

Constraint: The platform needed to analyze clothing images, automatically remove backgrounds, and provide personalized outfit recommendations with sub-second latency on mobile devices, all while operating under a strict cloud budget.

The Solution

How I Solved It

Trade-offs & Architecture: We opted for a microservices architecture deploying models via FastAPI on RunPod, rather than monolithic cloud ML services which were too expensive. I designed a pipeline prioritizing the background removal model first, cascading to styling models asynchronously to keep the primary UI responsive.

Results Achieved

Led the team as Team Lead across the full project lifecycle
Image analysis model deployed in production
Real-time background removal pipeline
Outfit recommendation engine live
Collaboration models for Exhibit Social App

Key Takeaways

1. Invest in real-time inference infrastructure early — latency kills user engagement in ML products.
2. Background removal and style transfer models require aggressive optimization to run affordably at scale.
3. Cross-functional team leadership is as important as technical architecture when shipping ML products.

Want results like these?

Let's discuss how I can help your project succeed.

Start a Project