Team
AllKnighters
Project Concept
No description has been added yet.
Entry
Status: Submitted
Last saved: May 11 at 2:41 AM PDT
Team Roster
Message board not available for this team yet.
Priyanshu Chawda Team Lead RSVP Approved
Student at PES Modern College of Engineering
Priyanshu Chawda led the full project end-to-end, including product concept, architecture, frontend implementation, agent workflow design, generative UI surfaces, MCP widget integration, testing, and final demo polish.
He designed AgentOps Flight Deck as a control tower for AI agent failures, focused on the hackathon theme of moving beyond chat into runtime-generated interfaces. He built the core AgentOps experience: mission/run state, failed tool-call timeline, evidence cards, risk gates, repair options, approval queue, raw trace view, and final report/score surfaces.
On the frontend, he used CopilotKit with React/Next.js to embed the agent experience inside the application and connect agent actions to live UI state. He implemented the generated UI pattern using a trusted component catalog inspired by A2UI principles, ensuring the agent renders structured dashboards and approval flows instead of unsafe arbitrary HTML.
He used AG-UI/CopilotKit-style state updates to make the frontend react to agent tool calls in real time, and used MCP Apps / mcp-use to expose AgentOps surfaces as reusable tool-backed UI widgets. He also integrated Google DeepMind Gemini for agent reasoning and LangGraph/LangChain-style orchestration for structured agent behavior.
For demo reliability, he added a deterministic Judge Mode and Playwright E2E tests so judges can see a complete working flow: a failed `browser_search` agent step becomes an interactive timeline, evidence panel, medium risk gate, repair option, approval queue, and report. He also handled safety constraints such as simulated-vs-real labeling, no fake evidence URLs, no external writes, and catalog-based rendering only.
Priyanshu Chawda , actively seeking knowledge sharing and co-founders for technical ventures. They are open to introductions and prefer to be contacted via email. Priyanshu is currently working on projects like YouTube Flashcards, an AI tool for educational content, and ghfind, a GitHub searching utility. Their expertise lies in building agentic systems and Model Context Protocol (MCP) servers using TypeScript and Python, with a focus on LLM orchestration, Gemini API integration, and full-stack AI applications for developer productivity and automated knowledge extraction. Priyanshu is also open to work and is interested in technical architecture and developer marketing.
Technical architecture, developer marketing, agentic systems, Model Context Protocol (MCP) servers, LLM orchestration, full-stack AI applications, automated knowledge extraction, developer productivity tools, TypeScript, Python, Gemini API integration.
Current projects include **YouTube Flashcards**, an AI tool for educational content, and **ghfind**, a utility for searching GitHub. He is actively building agentic systems and Model Context Protocol (MCP) servers using TypeScript and Python. His work involves LLM orchestration, integrating the Gemini API, and developing full-stack AI applications focused on developer productivity and automated knowledge extraction.