免费 · 约 6 分钟 · 免登录
为你真正需要的 AI 团队制定蓝图——按月、按预算、按地区。
诚实的自建 vs 外援建议："在伦敦，12 个月内你最多能现实地招到所需 5 名高级 AI 工程师中的 2 名。考虑前 6 个月外援 3 名。"覆盖 11 个地区的薪资带。六分钟得到 24 个月的计划。
角色
十一个核心角色。向导挑选你需要的。
AI 工程师
负责提示词、RAG、评估、集成。多数团队最早需要的 AI 招聘。
ML 工程师
训练并部署模型。与 AI 工程师不同——不同的循环、不同的薪资带。
MLOps / 平台
GPU 运维、可观测性、基础设施。让团队其余成员产出更高的角色。
数据工程师
管道、检索存储、数据基础设施。最难省略的角色。
AI 产品经理
AI 原生的产品感知、评估标准、路线图。一半的胜利来自决定不做什么。
AI 评估 / QA
评估、红队、测试基础设施。在用户发现之前捕捉模型回归的角色。
地区
贴合你实际情况的薪资带和招聘周期。
美国
深度人才密度，最高薪酬。SF Bay / NYC 在 2024 年后远程友好。
英国 / 伦敦
深度密度，IR35 + 税务复杂。强大的应用科学家人才池。
德国 / DACH
强大的 ML 工程能力。更高的负载倍数（1.45）。招聘缓慢但稳定。
印度
初级/中级深度密度，高级人才正在增长。招聘周期通常是全球最快的。
新加坡 / 阿联酋
中等密度。高级人才溢价。签证摩擦真实但可行。
全球远程
最广人才池，EOR + 税务复杂。5–9 个月的招聘周期是诚实的。
方法论
无介绍费。无供应商赞助。
薪资带在 Levels.fyi、Built-In、Payscale、Howdy、Second Talent、LinkedIn Talent Insights 和 HeroHunt 之间三角校验——每季度刷新。自建 vs 外援逻辑以伪代码公布，便于你压力测试。
阅读完整方法论常见问题
关于 AI 团队规划的常见问题。
What does the Buzzi.ai Team Blueprint tool do?
It produces a month-by-month AI team plan tailored to your project type, scale, region, timeline, and budget — with build-vs-augment recommendations, salary-banded budget envelope, and a JD-template pack you can paste into your ATS.
Why does the tool separate AI Engineer from ML Engineer?
Different jobs. AI Engineer owns prompts, RAG, evaluation, integration. ML Engineer trains and deploys models. Hiring loops, comp bands, and skill profiles diverge — collapsing them produces wrong plans.
Where do the salary bands come from?
Triangulated across Levels.fyi, Built-In, Payscale, Howdy (LATAM/India), Second Talent (APAC/MENA), LinkedIn Talent Insights, and HeroHunt time-to-hire data — refreshed quarterly.
What does "loaded multiplier" mean?
Cost-to-employer beyond base. Includes payroll taxes, benefits, equipment, and (region-specific) housing allowance, super, IR35 carve-outs, EOR fees. Defaults: US 1.30, UK 1.30, Germany 1.45, Singapore 1.30, Australia 1.25, India 1.20, UAE 1.20, Canada 1.25.
How does the build-vs-augment logic work?
Tight timeline + medium-density region + senior-or-above seniority → augment recommended for the bridge months. Build posture override respects your preference but still surfaces the gap.
When should I hire a junior vs a senior?
Pilot scale prefers junior-to-mid with a senior anchor. Mid and enterprise production scale need senior + staff layered, with leadership earlier (month 3-6).
When does augmentation win over hiring?
When time-to-hire exceeds half your timeline, when seniority is hard to source in-region, or when you need a deep skill for less than 6 months.
What's the typical time-to-hire in different regions?
Deep markets (US, UK, Germany, India): 2–4 months. Medium markets (UAE, Singapore, Australia, Canada, France, Netherlands): 3–6. Global-remote: 5–9 with EOR / tax complexity.
Are the JD templates authentic, or AI-generated boilerplate?
JD packs use a Buzzi-curated template per role with placeholders for your stack and AI-generated "About our AI work" blocks personalised to your inputs. Always reviewed by counsel before posting.
Do you cover regional labor laws?
Yes for the major contours: IR35 (UK), super (Australia), EOR / WIF (global), 13th-month pay (Germany / EU), DPDPA (India). The methodology page links to the authoritative sources.
How do remote-hiring tax complexities work?
Loaded multipliers reflect typical EOR fees (10–15%). For >5 remote hires, build a permanent establishment analysis — the methodology page includes the framework.
How does equity differ by stage?
Seed: 0.10–0.50% senior IC. Series A: 0.05–0.20%. Series B+: smaller, RSU-style at later stages. v1 reports band only — refined modelling planned.
What visa considerations apply?
H-1B (US), Skilled Worker (UK), Blue Card (EU), Green List (UAE), Tech.Pass (Singapore). Each adds 1–3 months to time-to-hire and constraints on seniority swaps.
How often are salary bands updated?
Quarterly refresh from primary sources, with hot-region overrides as labour markets move. Last verified date is shown in admin per band.
How confident are you in the salary data?
Source confidence scored 0.0–1.0 per band. Deep markets (US / UK / India) score 0.85+. Medium markets 0.65–0.80. Global remote 0.55–0.70. The methodology page calls out where uncertainty is highest.