Ajay Rana

• Sole owner of AI model training, RAG data ingestion, and Python backend development for a live SaaS AI Sales Agent handling 120K+ real-time queries. • Built end-to-end RAG pipelines with embeddings, vector storage, and semantic retrieval using FastAPI-based RESTful APIs. • Architected event-driven microservices backend with async task queues via Celery + Redis with backpressure and dead-letter queues. • Engineered LLM-powered extraction using structured prompting and Pydantic v2, reducing manual normalization by ~40%. • Managed MongoDB, Redis, and Elasticsearch for data storage, caching, and vector search. • Deployed via Docker and Jenkins CI/CD, driving 30%+ query throughput improvement.