Samip Ghimire

At Proventus Analytics, I shipped production AI agents, MCP tool integrations, voice-calling agents, and Python backend services for real business workflows. I own delivery end-to-end, from requirements and deployment through monitoring, debugging, and production fixes.

I build multi-step LLM systems with tool calling, RAG, LangChain, LangGraph, CrewAI, FastAPI, and structured outputs. My work includes connecting agents to external APIs, databases, files, and business systems while handling authentication, data flows, error recovery, testing, and reliability.

I focus deeply on AI/LLM security and red teaming, probing agentic systems against the OWASP Top 10 for LLM Applications. I use automated adversarial evaluation, prompt injection and jailbreak testing, guardrails, and structured output validation to harden systems before and after release.

Beyond production work, I've built RLVR-trained CAD/CAE code-generation agents, a diffusion-based virtual try-on system, incident-management and meeting-intelligence agents, and internal automation products. I prototype quickly, validate with real users, and iterate toward systems that remove manual work.