Software Development Engineer – Backend
ConvoZen.ai NoBroker's AI subsidiary
- Platform V2: helped re-architect the platform, consolidating 300+ Kafka topics into a clean routing system using CQRS, event sourcing and saga orchestration.
- Multi-tenant backend: built a unified Spring Boot backend that replaced per-client Django instances, cutting infrastructure cost by 70%.
- Call pipeline: designed a Kafka-based call-analysis pipeline on Kubernetes with autoscaling that processes 800K+ calls a day, with priority queues for time-sensitive clients.
- Job server: built a multithreaded job server with Spring Boot and Kafka; better consumer-group partitioning and parallel batches improved efficiency by 60%.
- Auth and security: built authentication from scratch with JWT, Redis-backed sessions, OAuth2 + PKCE and three-tier RBAC for enterprise accounts.
- AI and speech: multi-speaker transcription with Conformer models and Whisper fallback; summaries, sentiment and intent detection with Gemini and Azure OpenAI, cutting LLM API costs by 40% through batching.
- RAG products: a Customer Insights bot for querying call transcripts in natural language, and a RAG WhatsApp chatbot for the Odisha government.