Shiva Shrestha

I’m a Senior Software Engineer with 5+ years at Verisk, growing from trainee to technical lead across backend, cloud, and GenAI engineering. I design and ship LLM-powered authoring systems in the insurance domain, including RAG pipelines, real-time streaming APIs, and production AI guardrails for enterprise clients.

In my current role, I selected to design and ship Verisk’s first LLM-powered authoring platform for the insurance industry. I architect and deliver GenAI authoring APIs using Claude via AWS Bedrock and OpenAI, powering clause suggestion, form drafting, and automated content generation.

I focus on reliability and safety in production: I built production RAG pipelines over OpenSearch, engineered multi-layered LLM guardrails (prompt injection defense, PII redaction, content safety), and implemented real-time SSE streaming for low perceived latency. I also use prompt templates, system prompts, and evaluation harnesses to systematically optimize accuracy, latency, and cost.

Beyond engineering, I mentor teammates and help raise standards through code reviews and training—supporting Python trainees and leading technical interview assessments. I’m also an active educator (visiting faculty and instructor), and I’m seeking a GenAI/ML engineering role with an international team where I can apply my depth in building reliable, production-grade AI systems.