Neeraj d suvarna

I’m a Software Engineer specializing in backend development and AI-powered systems, with a strong focus on production-ready delivery end-to-end. I build with Python and modern API patterns, always thinking about performance, scalability, and reliability from day one.

At Moback Technologies India Private Ltd., I reduced TTS latency from 30s to 2–3s by replacing a GPU-heavy pipeline with Piper TTS (CPU). I also built an AI-driven interview engine using Core/Split/Blend/Hybrid strategies with resume-first weighting and multi-level difficulty, improving question relevance by 75% and evaluation accuracy by 30–35% using a 6-metrics scoring system.

Beyond models, I architect real systems: I designed a PostgreSQL + Supabase backend with RLS, auth-protected APIs, and 10+ Edge Functions. I integrated Dodo Payments webhooks/retries/retake logic and Mixpanel analytics (50+ events), then deployed on AWS & Vast.ai with Apache2 reverse proxy, SSL/TLS (Let’s Encrypt), DNS, and production-ready CI/CD pipelines.

I’ve also built multimodal RAG for enterprise knowledge assistants—creating a multi-tenant platform with secure, isolated retrieval and a ~30–40% improvement in data safety. In parallel, I developed resume screening pipelines (25–30% precision lift) and worked on ML for smart contract vulnerability detection on Ethereum, including 95%+ data quality and improved precision/recall/F1 by up to 12%.