AI Agentic Trading Expert

Job description

#### **Key Responsibilities** **Build the Capabilities Market (Supply Side)** - Design and ship the official skills spanning our 14-domain taxonomy (Leverage, Assets, Portfolio, Pulse, Plays, Risk, etc.) — covering market analysis, on-chain liquidity, identity resolution, derivatives, and DeFi primitives. - Define publishing standards, SDK ergonomics, and the skill-signing / revenue-share contract that third-party developers will build against. - Own the MCP server and CLI that make skill authoring feel as easy as writing a Vercel function. **Drive Agent-Side Integrations (Demand Side)** - Partner with teams to land CMC for Agent skills inside live agent products. - Implement A2A / cross-agent protocol adapters so a skill published once is discoverable and callable everywhere agents live. - Feed integration learnings back into the SDK — every painful integration becomes a docs page or a helper API. **Grow the Developer Community (Ecosystem & Hackathons)** - Co-design and run **CMC for Agent hackathons** — tracks, bounties, reference repos, judging rubrics, and the technical support Discord during event weekends. Deep collaboration expected with BNB Hack Local Series, YZi Labs EASY Residency, and Virtuals/ACP ecosystem events. - Write the technical content that makes developers choose us: launch posts, cookbook recipes, video walkthroughs, and "build-with-us" threads. - Be the technical face at Web3 × AI events; **Engineering & Grounding** - Improve LLM grounding on crypto-specific context (tokens, contracts, narratives) to cut hallucination and latency on trading-adjacent queries. - Build evaluation harnesses that measure skill accuracy, reliability, and economic honesty — not just latency and uptime. #### **Requirements** - **Crypto-native fluency**: you can explain DEX AMMs, perp funding, bridge risk, and on-chain identity without reaching for Google. Comfortable reading Solidity even if you don't write it daily. - **AI agent systems, hands-on**: function calling, tool use, MCP, and at least one of LangGraph / OpenClaw in production — not just tutorials. - **Developer-product instincts**: you've built or heavily contributed to an SDK, CLI, or API that outside developers actually adopted — and you have opinions about why most developer docs are bad.