| Surface | Role | Route |
|---|---|---|
| GitHub Profile | Public homepage and project facade | github.com/HaodiFan |
| AF-wiki Map | Fitness memory, knowledge graph, source boundary, and area navigation | haodifan.github.io/AF-wiki |
| AnthonyHF.Skill | Personal avatar skill matrix and Anthony-specific context router | haodifan.github.io/AnthonyHF.Skill |
| ChatGPT Next Web | App-layer AI chat deployment | chat-gpt-next-web-haodifan.vercel.app |
Most AI work fails somewhere between model capability and real-world use. My work is about closing that gap: understanding the ceiling, shaping the data frontier, and shipping systems that survive production.
| Layer | Focus | What I bring |
|---|---|---|
| Ceiling | NVIDIA Tegra SoCs, Xavier / Orin | Hardware-aware system thinking, profiling, and architecture tradeoffs |
| Frontier | Frontier-model data production | GUI Agent trajectories, QC workflow, and high-throughput data operations |
| Ground | Business delivery | 30+ AI solutions across 4 industries, from RPA to LLM and GUI agents |
MetaInFlow: AI delivery infrastructure for downstream markets.
The bottleneck is not only algorithms. It is delivery: turning frontier AI into maintainable, ROI-positive business systems.
2007 First line of code
2016 UIUC — Electrical & Computer Engineering
2018 NVIDIA Shanghai — Tegra SoC team
2021-2025 Shanghai Zhijue Information Technology Co., Ltd. — AI Tech Lead
2022-2025 MetaUniTech — personal brand and Explore AGI work
2025-2026 GrainedAI — Co-founder & CTO
2026 -> MetaInFlow — Co-founder & CTO
| Date | Area | Update |
|---|---|---|
| 04-20 | 🔧 System | area 层收敛为动态模块：knowledge 提升为 area，并补 work area 骨架 — → |
| 04-19 | 🔧 System | LeadFlow 架构正式命名，并统一仓库与主页文案 — → |
| 04-18 | 📋 Planning | W17 训练计划改为下肢、游泳、推拉混合，并匹配营养节奏 — → |
| 04-18 | 🏊 Fitness | 游泳 500m，配速 4'37/100m，平均心率 150 bpm — → |
| 04-18 | 🥩 Nutrition | 晚餐牛腱子加牛肉约 750g，当日蛋白基本已覆盖 — → |
| 04-18 | 📋 Planning | 当前减脂主计划固定为 3 次游泳加 3 次力量代谢训练 — → |
| Repo | Why it matters |
|---|---|
| AF-wiki | Public second-brain source for areas, fitness memory, knowledge curation, source manifests, and the Pages map. |
| AnthonyHF.Skill | Personal avatar skill matrix and Anthony-specific working context. |
| snapAF-skills | Local skill collection for media, productivity, domain, and devops workflows. |
中文版本
我关注的不是“AI 看起来很强”，而是它能不能在真实业务里稳定落地。我的工作长期围绕三层展开：硬件和系统边界、前沿数据生产、业务交付闭环。
| 层级 | 关注点 | 我的能力 |
|---|---|---|
| 天花板层 | NVIDIA Tegra SoC，Xavier / Orin | 了解硬件边界、性能约束和系统架构取舍 |
| 前沿层 | 前沿模型数据生产 | GUI Agent 轨迹、QC 流程、高吞吐数据生产 |
| 地面层 | 业务交付 | 4 个行业 30+ AI 方案，从 RPA 到 LLM / GUI Agent |
MetaInFlow：面向下游市场的 AI 交付基础设施。
真正的瓶颈不只是算法，而是交付：把前沿 AI 变成可运行、可维护、能产生 ROI 的业务系统。
| 日期 | 领域 | 动态 |
|---|---|---|
| 04-20 | 🔧 系统 | area 层收敛为动态模块：knowledge 提升为 area，并补 work area 骨架 — → |
| 04-19 | 🔧 系统 | LeadFlow 架构正式命名，并统一仓库与主页文案 — → |
| 04-18 | 📋 规划 | W17 训练计划改为下肢、游泳、推拉混合，并匹配营养节奏 — → |
| 04-18 | 🏊 健身 | 游泳 500m，配速 4'37/100m，平均心率 150 bpm — → |
| 04-18 | 🥩 营养 | 晚餐牛腱子加牛肉约 750g，当日蛋白基本已覆盖 — → |
| 04-18 | 📋 规划 | 当前减脂主计划固定为 3 次游泳加 3 次力量代谢训练 — → |