At Reqpedia, I take end-to-end engineering ownership of SnappySell, improving core workflows, reliability, and user experience. I built its pricing and unit-economics framework and designed face-blur preprocessing to protect source-model identity while preserving garment details.
I expanded video generation with a custom option and implemented Replicate as a fallback for core AI services. I also led backend development of the Platform Admin module for visibility into user activity, credit spend, adoption, and platform economics.
As an IT Intern at Reqpedia, I led R&D on generative AI model selection and authored a cost estimation document that guided an API-first architecture. I designed a normalized database schema with 10+ interconnected tables and delivered AI-generated avatar and model variants for the product.
At Culinda Inc., I optimized Java REST APIs and SQL queries, and refactored a reusable credit-based subscription workflow. I also independently scoped and built a full-stack HRMS platform; on Nova and CivicVoice, I contributed to AI voice-agent and civic grievance workflows.
