Architecting for Exponential Growth

Integrated AI-driven Machine Learning pipelines into the serverless architecture using AWS SageMaker, enabling real-time personalization at scale for individual user preferences and artistic interests. Implemented automated failover mechanisms with multi-region deployment, predictive scaling algorithms leveraging historical traffic patterns and seasonal trends, and adaptive caching strategies that reduced infrastructure complexity by 65%.

In parallel, architected Lift — an evolution of the Art Inspiration prototype — to solve the storytelling gap in the artist marketplace. Designed and built a web-based custom editor where curators compose animated presentations from video and image assets, with GSAP and Canvas delivering frame-accurate visual sequences in the browser. The backend — built in Python and PHP — processed each artwork through an asynchronous queue pipeline: media was ingested, narration scripts were translated via Google Translate into 10+ languages, and Azure Cognitive Services synthesised natural-sounding audio for each locale. The queue-driven architecture decoupled the editor experience from heavy processing, ensuring curators could publish without waiting on render jobs.

Extending the platform into physical spaces, delivered a gallery kiosk in 4 weeks — a self-contained Flutter application running on Raspberry Pi, driving both the display and a colour postcard printer. Visitors explored art forms through a custom internal Art Wiki API, then selected and printed a branded physical postcard as a keepsake, closing the loop between the digital marketplace and the real-world gallery experience.

To harden seller trust at the point of onboarding, engineered a real-time liveness verification system in Angular using TensorFlow.js in 8 days: the system randomly assigns 3 of 6 facial challenges — blink, smile, head turns and tilts — instructing users via bilingual voice synthesis in English and Hindi with on-screen text as fallback, capturing a verified image and video on successful completion entirely within the browser.

The platform achieved a 90% customer satisfaction score and saw a 60% surge in user engagement driven by hyper-personalization and seamless discovery. Lift brought 300+ artworks to life as immersive, narrated reels — accumulating 10,000+ views and opening international buyer segments that static listings could never reach.

The gallery kiosk extended Rooftop's brand into the physical world, turning gallery visits into interactive experiences and putting a branded postcard in every interested visitor's hand — a tangible marketing touchpoint delivered within a 4-week build cycle.

The liveness verification system established a scalable anti-spoofing layer for seller onboarding in just 8 days, with bilingual voice-guided challenges removing barriers for Hindi-speaking providers and eliminating manual review overhead entirely. Established itself as the go-to solution for art workshop and artwork discovery with zero critical incidents over 2 years of operation. Reduced infrastructure costs by 40% through intelligent resource optimization while simultaneously improving response times by 35% and enabling 10x growth without service degradation.