Hackathon Portal
AI Tinkerers - Hong Kong
Team

PengChauNerd

Project Concept

No description has been added yet.

Entry

Status: In Progress

Last saved: May 09 at 5:39 PM HKT

Team Roster

You must be registered for the event to view the team message board.

Adrian Leung Team Lead RSVP Approved

Product Manager at Votee
Adrian Leung (Lead — solo team) Designed, scoped, and built the entire stack end-to-end: • Product & scope — ran two rounds of Socratic deep-interview with Claude Opus to crystallize 3 generative-UI skills, 6 use cases, and a 2.5-hour build budget. Authored the 8-doc PRD pack in docs/gtm-intelligence/ (PRD, Use Cases, User Scenarios, System Architecture, ADRs, Implementation Plan, Demo Script, Risk & Plan B). • Frontend (Next.js 14 + Tailwind + CopilotKit) — built the canvas with three useFrontendTool handlers (render_quadrant_map / render_pitch_card / render_data_gap), the SessionSidebar with star + filter, html2canvas PNG export, and the Supabase-backed session resume flow. • Agent (Python LangGraph + Claude Sonnet) — wrote the GTM system prompt with explicit anti-hallucination routing, tool registry, and the live Beever Atlas MCP client wrapper (search-wiki + ask-atlas via fastmcp). • MCP layer (TypeScript, mcp-use) — implemented the two local tools (get_product_specs, get_pr_records) with Zod schemas + mock JSON. • Live Atlas integration — pre-seeded the Beever Atlas docker stack (MongoDB + Neo4j + Weaviate) with structured competitor docs in Phase -1 (night-before prep), wrote the parser that converts free-form Atlas content into the quadrant render schema. • Persistence (Supabase) — schema design (sessions, messages, artifacts), anonymous-user UUID flow via localStorage, RLS-disabled hackathon mode. • Pitch deck — built the 16-slide HTML deck under docs/gtm-intelligence/slides/ in white & blue theme, including the vibe-coding journey timeline.
I am a Product Manager at Votee AI with a foundation in Product Design from global firms like EY and Viu. Our coverage, my work focuses on driving product-market fit for AI solutions, spanning enterprise agentic systems. I specialize in bridging the gap between complex AI engineering and user-centric design to solve real-world problems. I am passionate about go-to-market strategies that empower cross-functional teams to deliver research-driven, impactful products that scale.
How can we use AI to move beyond static components toward interfaces that adapt to user intent in real-time? I’m keen to connect with AI researchers, product designers, and full-stack builders who are interested in leveraging LLMs for automated code generation, dynamic frontend building, and trying to break through the barrier of AI engineering, Product Design, and business development.
- Task-based Agent(Skills, Planner, Executor, Memory) - LLM Wiki(Channel-based Memory for enterprise knowledge base) - AI Agent chatbot projects - AI agentic compliance checking solution