Nirav

Key Skills

Software

Appen

CVAT

Labelbox

LabelImg

V7 Labs

Surge AI

Encord

Roboflow

Supervisely

Argilla

Sama

Hasty

Tagtog

Doccano

Data Annotation Tech

OneForma

Mindrift

Internal/Proprietary Tooling

Other

Top Subject Matter

Computer Vision Annotation

Top Data Types

Video

Computer Code Programming

Medical Dicom

Top Task Types

Segmentation

RLHF

Computer Programming/Coding

Function Calling

Data Collection

Company Overview

Xenveo: Specialized Physical AI Training Data Partner Overview Xenveo is a specialized AI training data company providing data collection, annotation, and curation services across multimodal, multilingual, and physical-world domains. We operate a workforce of approximately 1,000 annotators concentrated in India, with field collection capability across Brazil, Spain, Portugal, and France, and US/Canada client coverage. Xenveo is the sister company to Canterr, Inc., a US-based technical consulting and AI/cloud delivery firm. The two share leadership, infrastructure, and a talent-finds-talent operating model where engineers recruit engineers and annotators recruit annotators. This compounds workforce quality and trust without depending on paid acquisition. Mission Build the specialized data layer that frontier AI needs but commodity labeling marketplaces cannot easily deliver: physical-world capture, multilingual coverage across underrepresented languages, regulated-domain access (medical, robotics), and custom protocol design, all under documented consent, security, and quality frameworks. Fit with OpenTrain’s Marketplace OpenTrain's marketplace serves AI companies that need vetted talent across 190+ countries, with both self-service and managed offerings. Xenveo fits OpenTrain's model in two complementary ways: 1. As a vetted outsourced labeling provider on the OpenTrain marketplace bringing 1,000+ annotators with multilingual coverage and domain specialization. 2. As a capacity partner for managed-service engagements that exceed standard freelance fulfillment such as physical-world capture, medical imaging, multilingual non-English work, and other specialized pipelines. Both models complement OpenTrain's any tool, any domain, transparent pricing thesis. We bring workforce, capture infrastructure, and domain partnerships into your platform without competing with your marketplace flywheel. Services • Image, video, and audio collection at scale, including in-person and field capture • Annotation and labeling with sustained capacity of approximately 1M labels per month across 2D/3D bounding boxes, segmentation, keypoint, classification, transcription, and preference data • Multilingual data services in English, Spanish, Portuguese (LATAM and Iberian), Hindi, Gujarati, Tamil, and Telugu • Physical and in-person collection via rigged capture stations, on-location shoots, and crowdsourced field campaigns • Custom dataset design, capture protocols, annotation ontologies, QA pipelines, and consent frameworks tailored to buyer requirements • Tool-agnostic delivery, we work in client annotation environments (Label Studio, CVAT, Encord, Labelbox, Argilla, doccano, or proprietary platforms) and also operate in our own workflow tooling Specialized Verticals General multimodal data (in production today). Image, audio, video, and text data for foundation model training. Multilingual coverage including underrepresented Indian and LATAM languages. Physical AI training data (active build-out). Warehouse manipulation, humanoid robotics, embodied AI capture. Capture rig bills of materials documented at approximately $2,800–$3,000 per station. Annotation ontologies designed for kinematic, video, and imitation learning data. Medical imaging data (in development, partnerships forming). Hospital sourcing pathways established with several NABL-accredited diagnostic chains and NABH-certified hospitals with offering MRI, CT, PET-CT, SPECT-CT, mammography, echocardiography, pathology, and neurology imaging. Senior surgeons and medical personnel serve as Strategic Clinical Advisors, and drive high quality and efficiency in medical annotation. Federated-learning architecture under design. Surgical robotics and clinical workflow data (early stage). Five data tracks scoped: surgical video annotation, kinematic data, imitation learning, 3D imaging, and pre/intra/post-operative correlation. Agricultural robotics data (early stage). India and Asia field supply for agri-tech buyers. Unique Tools and Methods Annotators-recruiting-annotators workforce model. Talent grows through referrals, producing faster vetting and higher trust than commodity marketplaces. Capture rigs for physical AI. Documented station BOMs designed for warehouse manipulation, humanoid teleoperation, and embodied AI data capture. Standardized rigs enable repeatable quality across sessions and locations. Federated learning architecture (in development). For medical imaging, we are building a federated-learning-primary pipeline via NVIDIA FLARE, with a premium static-delivery tier where data egress is permitted under contract. Dual-track consent regime under design. In-house workflow platform. Frontend: React with TanStack Start (hybrid SSR plus client). Backend: Nest.js (TypeScript), Prisma ORM, Bun runtime, Postgres 17. Used for task assignment, QA review, and reporting. Tool-agnostic, we deploy annotators into client environments by default. Field campaign infrastructure. Payment rails operational via Wise Business for EU contributors and PIX-native providers (e.g., Transfero) for Brazil. GDPR and PII protocols documented. Active campaign scoping includes crowdsourced barcoded document collection across Brazil, Spain, Portugal, and France. Workforce • Approximately 1,000 annotators across various specializations, in India, Brazil, and partner regions • Native-language coverage in English, Spanish, Portuguese, Hindi, Gujarati, Tamil, and Telugu Locations • Annotation and operational base: India • Field collection footprint: Brazil, Spain, Portugal, France • Client coverage: United States and Canada Security and Compliance • ISO certified (Canterr, parent infrastructure) • SOC 2 in progress • Documented consent framework for all data subjects • Federated learning architecture under development for sensitive clinical data, data stays at source, models travel • GDPR and PII protocols documented for EU field campaigns Client References References available under NDA on request.

Security