Hashwanth Sutharapu I ship agents that actually run in production. Not demos, not notebooks, real systems handling 1000+ requests per minute. I'm an Agentic AI Engineer at MAQ Software (a Microsoft Partner), building multi-agent systems for Microsoft. My work sits at the orchestration layer: LangGraph/Microsoft Agent Framework for agent control flow, MCP for tool access, Azure AI Foundry and Copilot Studio for deployment, and a full eval and observability stack underneath. MS in Computer Science from Arizona State University (GPA 3.93). Some recent proof: an NL2SQL agent serving 500+ business users (25% reduction in cognitive load), a published MCP server running at 1000+ req/min with 99.9% uptime across 3+ teams, ThinkNet (an 8-agent deep researcher on LangGraph and AWS Bedrock), and a provisional patent filed on AEPS, an agent protocol stack for intra-org agent identity. What I'm building toward: context engineering as a first-class discipline. KnowledgeMCP, my current side project, cuts agent context retrieval from 3-4K noisy tokens down to 500 grounded tokens via hybrid search and RAG, the kind of upstream work that makes inference actually cheap and actually reliable. Agentic AI · MCP · Harness &
Software Data Operations Engineer
Software Data Operations Engineer Intern
Software Engineer 1
Associate Software Engineer
Masters, Software
Bachelors, Software
