Frequently asked questions.

Teehoo AI builds the human infrastructure layer for frontier AI labs — embodied AI data for humanoid robotics and expert evaluation for frontier LLMs. We sit between AI labs and the workforce that produces ground truth.

Teehoo AI is a global team currently based in Los Angeles, with operations across five offices in the US, Singapore, and China. We operate as a unified organization with shared leadership and product roadmap across all locations.

Five offices across three jurisdictions: Los Angeles (Global HQ), Singapore (APAC hub), Beijing (China HQ), Anhui (operations center), Jiangxi (embodied AI center). Plus a distributed multi-city workforce network across 10+ Chinese cities.

Three things: (1) tier-gated experts, not generic crowdworkers — we don't compete on commodity labeling economics; (2) compliance-native delivery posture — data sovereignty and in-region delivery for engagements that require it; (3) verification-first thesis — purpose-built for post-training evaluation, not retrofitted from data-labeling roots.

We disclose investors and advisors as engagements formalize. We don't list placeholder investor logos.

Six layers: Reasoning Evaluation, Agent Verification, RLHF Pipelines, Expert Domain Data, Embodied AI Data, and AI Safety & Red Team. Each is methodology-led with tier-gated experts, multi-judge agreement, and continuous calibration loops.

No. We align with public benchmarks (SuperCLUE, C-Eval, CMMLU, MMLU, SWE-bench, GAIA, HumanEval, FLAMES) and extend bilingual coverage on top — but we don't publish proprietary benchmarks that lock customers into our scoring. See research →

Real-world scene collection (homes / offices / factories / warehouses / transport), VR-assisted capture, first-person interaction recording, human motion simulation, and robotics manipulation data. Operated from our Jiangxi embodied-AI center with dedicated collection sites.

Any progress bars and capability matrices currently on our site reflect capability coverage scope — what we can measure — not real model scores. We will not publish model evaluation numbers unless they are real, reproducible, and contextual. Capability scope is disclosed as capability scope; traction is disclosed as traction.

Most engagements begin with a 2-week pilot: 50-100 workforce deployment, scoped evaluation rubrics, QA + compliance integration, dedicated client delivery team. Pilots that succeed scale into multi-quarter contracts.

We deploy operational capacity in days, not months — distributed across multiple cities. Rapid 50 → 500 worker scaling is supported via the China multi-city network and Singapore APAC hub. Specific scale ranges are scoped per engagement.

Yes. Enterprises deploying AI products and governments with high-stakes evaluation needs are both core customer segments — our compliance posture is purpose-built for these use cases.

Yes. Our multi-jurisdiction operations let us bind data residency to a specific region — China-only engagements stay on China cloud infrastructure; global engagements run with the geographic posture you specify per contract.

Go to the Expert Hub and apply. We review every application, route candidates to the domains where their experience fits, and respond on a rolling basis.

Six tier-gated tracks: STEM reasoning, agent & tool-use verification, finance / quant, medical & biomedical, AI safety & red team, and multilingual evaluation. Each track has its own rubric and tier system.

Real compensation tied to real production — never speculative, never delayed. Specific rates are disclosed per task / domain when you apply. We do not race-to-the-bottom on commodity labeling rates.

Most expert work is remote-friendly. Embodied AI tasks (VR / robotics capture) require on-site work at our collection centers; everything else runs on distributed remote operations.

No. Per our Cohort Routing policy, US-frontier-lab and international engagements route exclusively to global Chinese-speaking professionals based outside mainland China (H-1B / Green Card holders, international PR holders, overseas-resident specialists). China-resident experts route to China-domestic AI lab clients only through our in-region operations. This boundary is non-negotiable — it protects experts (export control compliance, data sovereignty), clients (regulatory clarity), and the integrity of the platform. Cross-jurisdiction routing is never default and never possible under any model (marketplace / job board / referral / sub-contract).

Engagements requiring data residency in China are routed through our in-region operations with in-region storage (Tencent Cloud China region), in-region compute, and regulator-aligned audit trails. Global engagements run on infrastructure scoped to the contract's geography.

Certifications are in progress, not yet awarded. We disclose certification status accurately — we will never claim certifications we haven't earned. Specific architectural alignment with each framework (data isolation, access tiering, audit trails, encryption) is in place at design level.

Engagements with cross-border component are routed through Singapore (APAC hub) with explicit contractual scope. Engagements requiring strict no-cross-border posture stay entirely within one jurisdiction.

Yes. We selectively partner with AI workforce platforms where the engagement complements our verification thesis — not commodity labor aggregation. Reach out to [email protected].

Yes. Aligned advisors (academic, regulatory, industry) and investors who understand the verification thesis are welcome — [email protected].

[email protected]. We respond within 3 business days.