Original posts on RAG systems, FastAPI, PostgreSQL, software licensing, and AI engineering — written by Diwash Bhandari, backend & AI systems engineer based in Kathmandu, available globally.
How to move a RAG prototype from demo to production: LangGraph ingestion, vector store selection, multi-provider LLM routing with LiteLLM, PII redaction with Presidio, and automated quality gates with Ragas and DeepEval.
OFFSET pagination dies at scale. Learn how keyset pagination (seek method) keeps PostgreSQL queries fast on tables with millions of rows, with a FastAPI + SQLAlchemy async implementation and real benchmarks.
How to build software licenses that are cryptographically signed, verify without calling a server, and support both node-locked and floating models. Ed25519 signatures, machine fingerprinting, offline grace periods.
WhatsApp CRM knowledge search that never confuses records: exact number lookup, pgvector semantic search, Postgres full-text search, and pg_trgm typo tolerance, merged into one answer with a leak guard.
I build RAG systems, FastAPI backends, AI agents, and licensing systems for teams worldwide. Let’s discuss your project.
Get in touch