Raj Simkhada

I’m an AI engineer focused on building production-grade machine learning and AI systems that turn unstructured information into structured outputs. In my current role, I work on backend services that power LLM workflows and customer-facing automation, with an emphasis on reliability and measurable improvements.

I’ve designed and built FastAPI microservices for AI workflows such as query routing, RAG-based Q&A, and secure email processing using API-key authentication. I also developed a production system that converts unstructured customer emails, FAQs, itineraries, and booking data into structured outputs—reducing manual response effort by 60%.

I care about performance and evaluation, so I’ve implemented retrieval improvements and operational safeguards. By using async processing, caching layers like Redis, and optimized vector search, I improved retrieval accuracy by 25% and reduced latency by 35%, while building RAG evaluation pipelines with metrics like Recall@K, Precision, and F1-score.

My work blends NLP, automation, and research-minded development. I’m actively publishing research on parameter-efficient fine-tuning (LoRA) for domain-specific recognition, and I’ve built projects ranging from agentic tour planning with stateful LangGraph services to domain-aware RAG assistants and interpretable prediction pipelines.