SadiGroup

SadiGroup Building the human intelligence behind AI.

SadiGroup delivers AI data collection, annotation, RLHF, speech datasets and multilingual services across 300+ languages worldwide.

πŸ‡¬πŸ‡· Greek transcription opportunity with SadiGroup!We are looking for experienced Greek transcription vendors and teams f...
10/09/2026

πŸ‡¬πŸ‡· Greek transcription opportunity with SadiGroup!

We are looking for experienced Greek transcription vendors and teams for a 2,000-hour audio project on DataPlus.

πŸ’΅ Rate: USD 45 per accepted valid audio hour, including transcription and QA.

πŸ“© Interested? Send your transcribers’ (TR) and QA reviewers’ email addresses through a private message to arrange the test.

Know a qualified team? Share this post or tag them below!

SadiGroup is building a network of native-language specialists for a potential remote AI evaluation project.We welcome a...
09/28/2026

SadiGroup is building a network of native-language specialists for a potential remote AI evaluation project.

We welcome applications from qualified professionals and vendors who can provide experts in healthcare, finance or law across these markets:

β€’ Mexico β€” Spanish
β€’ Indonesia β€” Indonesian
β€’ United Kingdom β€” English
β€’ Malaysia β€” Malay
β€’ Brazil β€” Portuguese
β€’ Thailand β€” Thai
β€’ Vietnam β€” Vietnamese
β€’ Saudi Arabia, Kuwait, Qatar and the UAE β€” Arabic

The work involves reviewing a question and four AI-generated answers, ranking the answers, and writing an accurate reference response.

If you have relevant professional expertise and native-language proficiency, please message us with your domain, native language, country of residence and CV. Vendors may share the same information for their available specialists.

This project is in the candidate screening stage. We will share task guidelines and trial details with suitable candidates before participation. CVs submitted for client review will have names, contact details and other personal identifiers removed.

SadiGroup is seeking experienced vendor teams in the United States and United Kingdom for a natural-conversation speech ...
09/28/2026

SadiGroup is seeking experienced vendor teams in the United States and United Kingdom for a natural-conversation speech data project.

Volume: 25 accepted valid hours of American English and 25 hours of British English
Vendor rate: USD 150 per accepted valid hour, all-inclusive
Scope: Speaker recruitment and compensation, recording, collection QA, annotation and annotation QA
Timeline: Approximately 2–4 weeks, subject to pilot approval

Recordings will cover transit, shopping, dining, office and other everyday environments. Each speaker may contribute up to 30 valid minutes. The pilot will use the Appen Task app and wired earphones.

Vendors only: DM SadiGroup with your US or UK coverage, available speaker capacity and relevant speech collection experience.

SadiGroup is seeking experienced vendor teams for two upcoming AI data projects:1. Autonomous driving: 3D LiDAR annotati...
09/25/2026

SadiGroup is seeking experienced vendor teams for two upcoming AI data projects:

1. Autonomous driving: 3D LiDAR annotation

We need teams experienced in 3D bounding boxes for vehicles, pedestrians and other road users; object classification; continuous-frame ID tracking; and QA across complex road scenes. Experience with sparse point clouds, occlusion, box alignment and yaw accuracy is essential.

2. Embodied AI: binocular first-person video collection

We need teams that can organize head-mounted binocular RGB recording with actual workers performing natural tasks in real workplaces. Priority environments include food service, retail, warehousing and logistics, automotive repair, and cleaning. Teams must be able to manage worker consent, recording quality and delivery across a diverse range of environments.

This opportunity is for established teams and vendors with relevant delivery experience.

To be considered, please send:

β€’ A brief company or team profile
β€’ Redacted screenshots, samples or case studies from similar projects
β€’ Your available team size, locations and realistic weekly capacity
β€’ Your QA process and indicative pricing
β€’ For video collection: available workplace categories and binocular camera capability

Please redact any confidential client or project information before sharing materials.

Send your profile and supporting evidence to [email protected] with the subject line β€œLiDAR Annotation Partner” or β€œBinocular Collection Partner.”

πŸ“£ Vendor partners wanted β€” 29,000 handwritten OCR images across 5 languages.SadiGroup is sourcing collection partners fo...
09/20/2026

πŸ“£ Vendor partners wanted β€” 29,000 handwritten OCR images across 5 languages.

SadiGroup is sourcing collection partners for handwritten OCR datasets in Hindi, Thai, Italian, English and German. We're looking for teams with genuine native-speaker reach in each market.

VOLUMES
β€’ Hindi β€” 6,000 images
β€’ Thai β€” 5,000 images
β€’ English (US / UK / Canada etc.) β€” 12,000 images
β€’ German (Germany) β€” 3,000 images
β€’ Italian β€” 3,000 images

REQUIREMENTS
β€’ All handwriting must be produced by native speakers in-country.
β€’ Sourced online: images must come from local native-language sites. The source URL must be retained for every image and delivered in a matching data sheet β€” this is checked at acceptance.
β€’ Collected on-site: local authenticity must be evidenced. Accepted proof includes photos of the writing environment, EXIF metadata showing collection location, or equivalent.
β€’ 1080p or above, clear and legible handwriting. Clean backgrounds with minimal distracting text preferred.

RATES (per image, online / on-site)
β€’ Hindi β€” $0.03 / $0.05
β€’ Thai β€” $0.03 / $0.05
β€’ English β€” $0.03 / $0.05
β€’ German β€” $0.03 / $0.09
β€’ Italian β€” $0.04 / $0.09

TO BE CONSIDERED, SEND US:
1. Your capacity per language β€” how many images you can realistically deliver.
2. Whether you'll provide online-sourced data, on-site collection, or both.
3. Your turnaround time and a sample of 10–20 images.

πŸ“© [[email protected]] β€” subject line: "Handwritten OCR β€” [language]"
Submissions close [22/09/2026].

πŸ“ New project β€” Tunisia & Libya | License Plate Image CollectionSadiGroup is sourcing at least 30,000 license plate imag...
09/18/2026

πŸ“ New project β€” Tunisia & Libya | License Plate Image Collection

SadiGroup is sourcing at least 30,000 license plate images each from Tunisia and Libya for an upcoming computer vision dataset.

What we need:

β€’ Real-world captures only β€” vehicles on roads or in parking lots. No staged or stock images.

β€’ Fully legible plates β€” the entire plate readable to the human eye. Colour JPG, minimal compression artifacts.

β€’ Deliberately varied plate pixel width:
– Small / distant: under 60 px
– Medium: 60–90 px
– Large / close-up: over 90 px
Small, distant plates are the hardest recognition scenario and our biggest gap β€” that's where we need the most volume.

β€’ Scene diversity β€” daytime and nighttime, frontal and diagonal angles, different weather and lighting conditions.

β€’ Plate type coverage β€” all currently circulating styles and number ranges: standard civilian plates plus common variants (commercial, temporary, regional/provincial).

β€’ Sample diversity β€” different vehicles across different regions and cities. No repeated shots of the same few plates.

Rate: $0.03 per image.

If you'd like to take part, tell us:
1. Can you capture in person/offline, or would you be using web scraping?
2. How many images can you realistically deliver for Tunisia and for Libya respectively?

Comment below or send us a message.

πŸŽ₯ We're Hiring Partners: Egocentric Video Data Collection in India, Japan & VietnamSadiGroup is launching a large-scale ...
09/17/2026

πŸŽ₯ We're Hiring Partners: Egocentric Video Data Collection in India, Japan & Vietnam

SadiGroup is launching a large-scale first-person (ego-centric) video data project to power the next generation of LLMs and Embodied AI, and we're looking for ex*****on partners, local studios, and fieldwork networks to join us.

πŸ“ Locations: India, Japan & Vietnam

πŸ“Œ Project Scope
⏱️ Volume: 2,000 hours total, with a minimum of 500 hours in Week 1
πŸŽ₯ Hardware: Provided by us (USB-powered, supports continuous 8–10 hour sessions)
πŸ“€ Data Upload: Direct submission through dedicated joint QA software

πŸ› οΈ Target Scenarios (Medium to Hard Tasks)
πŸ”¬ Labs & Metrology: calibration, titration, sample prep, weighing
πŸ’ Craft Workshops: jewelry setting, leather sewing, pottery, welding
πŸ–¨οΈ Print & Packaging: screen printing, vinyl wrapping, ink mixing
🍳 Commercial Kitchens & Maintenance: precision cooking, HVAC and equipment repair

βœ… Who we're looking for:
β€’ Studios, vendors, or field teams with access to real working environments like labs, workshops, print shops, and commercial kitchens
β€’ Capacity to scale fast and deliver high volume from the first week
β€’ A strong commitment to data quality and participant consent

⚑ Timelines are tight, so if you can mobilize quickly, we want to hear from you.

πŸ“© Comment below or send us a DM

Tag anyone in your network who runs a workshop, lab, or field team in these countries! πŸ‘‡

πŸ‡―πŸ‡΅ We're hiring: Japanese Historical Text Specialists (Remote)SadiGroup is looking for 2 native Japanese experts to work...
09/17/2026

πŸ‡―πŸ‡΅ We're hiring: Japanese Historical Text Specialists (Remote)

SadiGroup is looking for 2 native Japanese experts to work on a project involving modern-era Japanese history books and documents.

πŸ“Œ The work
β€’ Compare document images with OCR text and correct recognition errors
β€’ Translate the corrected text into natural modern Japanese

πŸ’° Pay
β€’ $0.10 per character
β€’ About 12,500 characters per person, around $1,250 per specialist

πŸ“… Timeline
β€’ September 2026 to February 26, 2027
β€’ Full-time, or part-time with stable weekly hours
β€’ 100% remote

βœ… Requirements
β€’ Native Japanese speaker
β€’ Degree in Japanese history, classical Japanese literature, or Japanese literary history (certificate required), OR professional experience reading historical documents at an archive, museum, or research institution

⭐ Nice to have
β€’ Experience with classical written Japanese (ζ–‡θͺžδ½“, ε€™ζ–‡) or kanbun
β€’ Experience with transcription (翻刻) or modern Japanese translation (現代θͺžθ¨³)

πŸ§ͺ Selection
1. Send your CV and degree certificate
2. Complete a short paid test (about 500 characters)
3. Onboarding and start

πŸ“© How to apply
Send a message to or email [[email protected]] with the subject "Japanese Historical Text Specialist".

Know someone with this background? Please tag them or share this post πŸ™

#叀文書 #翻刻 #ε›½ζ–‡ε­¦ #ζ—₯本史

VENDOR OPPORTUNITY | PROFESSIONAL RECORDING STUDIOS REQUIREDSadiGroup is seeking established recording studios or compan...
09/12/2026

VENDOR OPPORTUNITY | PROFESSIONAL RECORDING STUDIOS REQUIRED

SadiGroup is seeking established recording studios or companies with professional studio facilities to manage participant recruitment, onboarding, recording, audio-quality control and final delivery for three full-duplex natural-conversation projects.

Companies and professional studios only. We are not accepting applications from individual speakers.

PROJECT 1: CANADIAN FRENCH

Target: 150 hours
Participants: Native French speakers from Canada
Quantity: 70 pairs, with two participants per pair
Recording duration: Up to 2 valid hours per pair
Rate: USD 130 per pair per accepted valid hour
Total per pair: Up to USD 260 for 2 accepted hours

PROJECT 2: GERMAN

Target: 230 hours
Participants: Native German speakers from Germany
Quantity: 100 pairs, with two participants per pair
Recording duration: Up to 2 valid hours per pair
Rate: USD 140 per pair per accepted valid hour
Total per pair: Up to USD 280 for 2 accepted hours

PROJECT 3: SPAIN SPANISH

Target: 280 raw hours
Participants: Native Spanish speakers from Spain
Quantity: 50 pairs, with two participants per pair
Recording duration: 2.5 raw hours per pair
Rate: USD 110 per pair per raw recording hour
Total per pair: USD 275 for 2.5 raw hours
Deadline: October 20, 2026

RECORDING REQUIREMENTS

β€’ Participants must be 18-60 years old.

β€’ An overall 1:1 male-to-female balance is required, with a maximum fluctuation of Β±5%.

β€’ Regional accent diversity is required. Recordings must not be concentrated in only one or two regions.

β€’ Each participant may take part only once.

β€’ Each pair must record simultaneously in two separate, quiet and acoustically isolated rooms.

β€’ Required equipment: professional microphone, audio interface or sound card, computer and mobile phone.

β€’ Reference equipment: sE2200 microphone and Focusrite 2i2 audio interface.

β€’ Microphone audio format: mono WAV, 48 kHz, 32-bit.

β€’ Each speaker must be recorded on a separate track without crosstalk. The two tracks must be aligned within 0.1 seconds.

β€’ Two mono speaker tracks and one merged reference file must be delivered for each conversation segment.

β€’ Synchronized mobile-phone recording at a distance of 1, 3 or 5 metres may also be required.

β€’ The recording environment must be indoors, echo-free and at or below 50 dB.

β€’ Audio containing clipping, electrical interference, DC offset, continuous background noise, other voices or sudden noises will be rejected.

β€’ Conversations must be natural and spontaneous. Reading scripts or deliberately acting out conversations is not permitted.

β€’ Conversation topics will be provided. Each topic must be recorded as a separate segment lasting between 5 and 30 minutes.

β€’ Natural interruptions and overlapping speech should exceed 5%. Both speakers must participate meaningfully.

β€’ Violent, pornographic, politically sensitive or private content is prohibited.

β€’ Studios must manage participant registration, voice authorization, electronic consent, speaker deduplication, metadata collection, file naming and secure delivery.

TO APPLY

Please provide:

β€’ Company profile and website
β€’ Relevant voice-recording experience
β€’ Studio locations and photos
β€’ Complete recording-equipment list
β€’ Number of available two-room recording setups
β€’ Confirmed participant capacity by country
β€’ Expected weekly production capacity
β€’ Recruitment and onboarding process
β€’ Audio-quality control process
β€’ Earliest pilot date and final delivery schedule
β€’ Confirmation that you accept the stated rate

Email: [email protected]
Subject: Full-Duplex Recording Studio Application - [Country]

Building the Human Intelligence Behind AI.

Vendor Hiring | 1,000-Hour Full-Duplex U.S. English Voice Collection & AnnotationSadiGroup invites qualified U.S. data-c...
09/08/2026

Vendor Hiring | 1,000-Hour Full-Duplex U.S. English Voice Collection & Annotation

SadiGroup invites qualified U.S. data-collection companies, studios and managed field teams. Vendor applications only.

Scope
β€’ 1,000 accepted valid hours; at least 1,500 native U.S. English speakers (750 pairs)
β€’ 1.5 valid hours per pair; maximum 2 raw hours
β€’ Ages 18–60; 1:1 male/female ratio (Β±5%); one submission per person
β€’ Collection, transcription, annotation and QA

Regional allocation
β€’ New England - 100h: Boston; Portland/Providence
β€’ Mid-Atlantic - 200h: New York; Philadelphia/Baltimore
β€’ Inland North - 150h: Chicago; Cleveland/Detroit
β€’ Midland - 150h: Kansas City; Columbus/Indianapolis
β€’ South - 200h: Atlanta; Dallas/Nashville/Birmingham
β€’ West - 200h: Los Angeles; San Francisco/Seattle/Phoenix

Recording
β€’ Each pair records in separate quiet/soundproof rooms on synchronized channels with no crosstalk
β€’ Focusrite 2i2 interface and sE2200 microphones required
β€’ Microphone: WAV, 48kHz, 32-bit, mono
β€’ When assigned, far-field phone at 1m, 3m or 5m: WAV, 24kHz, 16-bit, mono; Android/iOS mix TBC
β€’ No echo; background noise ≀50dB; no clipping, truncation, DC offset or electrical interference
β€’ Separate 5–30-minute files per topic

Conversation
β€’ Natural, fluent, unscripted dialogue using supplied topics; no reading or imitation
β€’ Locally representative accents and clear speech
β€’ Natural overlap/interruption/competitive turn-taking above 5%
β€’ Both speakers must contribute meaningfully
β€’ No violent, pornographic, politically sensitive or prohibited content

Operations and compliance
β€’ Register gender, age, birthplace, upbringing location and device details
β€’ Obtain recorded voice authorization and signed electronic consent
β€’ Run identity and voiceprint de-duplication; duplicate pairs are rejected
β€’ Maintain consistent participant/group IDs, naming and metadata
β€’ Two mono tracks must align within 0.1s; a merged stereo file is also required per topic
β€’ Segment by sentence (average 5–6s; generally ≀8s), label speaker/gender, transcribe verbatim English and apply required noise, non-speech, overlap, laughter, outside-voice and personal-information tags
β€’ QA: β‰₯98% word accuracy, β‰₯97% speaker-role accuracy and β‰₯90% tag accuracy
β€’ Check waveforms/peaks, file size and MD5 uniqueness; deliver complete WAV, TXT and metadata V1.2 packages in the last week of each month, then a final consolidated delivery

Apply with:

1. Speaker capacity by region and weekly output
2. Studio locations and number of parallel rooms
3. Equipment/specification confirmation
4. Recruitment, recording, annotation and QA capability
5. Relevant project experience
6. Quote per accepted valid hour and/or participant
7. Mobilization timeline

Send via LinkedIn or [email protected]
Subject: US English 1,000H - Vendor Application - [Company]

SadiGroup
Building the Human Intelligence Behind AI

Address

Stella Court
Sunnyvale, CA

Opening Hours

Monday 9am - 5pm
Tuesday 9am - 5pm
Wednesday 9am - 5pm
Thursday 9am - 5pm
Friday 9am - 5pm

Telephone

+96176827310

Website

Alerts

Be the first to know and let us send you an email when SadiGroup posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Contact The Business

Send a message to SadiGroup:

Shortcuts

Share

Category