About Us · Mission & Vision
Ushering in a new wave of competent AI trainers to work for you.
AI models learn from human judgment. We think human evaluators deserve dignified pay, accessible tooling, and rigorous calibration — giving account owners and AI labs a verified, high-performance fleet.
Strategic Horizon
Our Vision
Elite Domain Competency
Crowdwork and low-quality data scraping have hit a ceiling. Frontier reasoning models demand subject-matter specialists — mathematicians, software engineers, linguists, and legal scholars — calibrated to evaluate subtle chain-of-thought arguments without hallucination.
Radical Economic Transparency
Taskers receive at least 30% of gross earnings directly with zero hidden kickbacks. Account owners retain 100% legal ownership and receive clean monthly balance sheets. We take a transparent 15% operational fee. No surprises, ever.
Zero-Trust Operational Armor
No passwords sent over web forms. All credentials route via Telegram's end-to-end encrypted secret chats. Tasking occurs exclusively in locked-down remote desktop environments with clipboard isolation and dedicated residential proxies.
611 Managed Accounts · 99.4% Platform Health · 4.9/5.0 QA Score
Operated strictly within terms on platforms that explicitly authorize managed contracting.
Guiding Principles
Our Operating Values
Compliance first
If a platform does not allow subcontracting, we turn the account away. Zero gray-hat workarounds.Open ledger pricing
Every split is published. Account owners and taskers see the exact math before any work begins.Guaranteed pay floor
Taskers always receive at least 30% of task earnings, paid weekly without delays.Universal accessibility
Asynchronous tasking designed for disabled, neurodivergent, and remote workers from day one.Cryptographic privacy
Passcodes and sensitive logins are handled strictly through Telegram's encrypted channels.Human dignity
We replace high-churn crowdwork with long-term specialist careers and direct admin support.
Specialized Operations
Our Leadership Team
Marcus Vance
Founder & Managing Director
Platform partnerships, regulatory compliance & enterprise staffing architectures
Pioneered verified human-in-the-loop workflows across major AI training platforms. Dedicated to closing the gap between idle accounts and competent talent.
Dr. Elena Rostova
Chief Learning Officer & Head of Trainer Development
Specialist trainer onboarding, multi-turn reasoning rubrics & cognitive ergonomics
Former researcher in computational linguistics and cognitive ergonomics. Authors our multi-step chain-of-thought training modules and domain qualification tracks.
Aris Thorne
Head of AI Quality Assurance & Evaluation Fleet
Multi-modal scoring rubrics, automated audit sampling & fleet scorecards (4.9/5.0 target)
Over 8 years designing rigorous evaluation pipelines for high-stakes ML systems. Manages our independent blind-sampling audits and quality appeal tribunals.
Kaelen Diaz
Director of Systems & Security Engineering
Isolated VM sandboxes, zero-clipboard confinement & Telegram encrypted dispatch
Information security veteran with a focus on containment architecture. Ensures host account credentials never touch public internet forms or unmonitored endpoints.
Sora Tanaka
VP of Contributor Relations & Accessibility
24/7 Telegram customer care, assistive technology accommodations & fair pay floors
Advocate for neurodivergent and disabled digital workers. Champions our non-negotiable 30%+ tasker pay floor and 24/7 real-time Telegram response desk.
Join Our Trainer Fleet
Are you a mathematician, software engineer, translator, or multimodal domain expert? We provide vetted accounts, hourly guarantees, and weekly direct payouts.
Frontier Methodologies
Industry Ecosystem & Benchmark Alignment
Nous Research
nousresearch.comRenowned creators of the open Hermes-series models, world-class synthetic reasoning datasets, and agentic benchmark datasets.
We calibrate our taskers on multi-step reasoning validation and chain-of-thought error verification standards established by Nous Research architectures.
EleutherAI
eleuther.aiPioneering non-profit AI research lab behind the standard LM Evaluation Harness, Pythia interpretability suite, and open alignment studies.
Our internal QA scoring rubrics incorporate standardized bias metrics and adversarial prompt testing frameworks inspired by EleutherAI research.
CarperAI (LAION Ecosystem)
carper.aiGrassroots research lab dedicated to advancing open-source Reinforcement Learning from Human Feedback (RLHF) and fine-tuning datasets.
Taskers are trained directly on multi-attribute preference ranking, pairwise comparison matrices, and reward model verification protocols.
Matthew Berman
youtube.com/@MatthewBermanOne of the most trusted independent AI educators and technical evaluators analyzing local LLM performance, open weights, and autonomous tools.
We integrate weekly stress-test analyses and practical agent evaluation frameworks highlighted in Matthew Berman's research into our continuing education briefings.
Yannic Kilcher / AI Pub
youtube.com/@YannicKilcherInfluential machine learning researcher and creator known for deep-dive paper critiques, adversarial prompt testing, and AI safety verification.
Our senior quality reviewers adopt red-teaming methodologies and critical paper breakdown rubrics popularized by Yannic Kilcher's technical breakdowns.
Continuous Rubric Evolution
As reasoning models (o1, R1, Hermes, Claude 3.7) introduce extended test-time compute, our taskers receive weekly briefings on emerging hallucination vectors, prompt drifts, and synthetic dataset decontamination.
Infrastructure & Tooling
Our Supporters & Technology Enablers
Multilingual Infrastructure Enabler
Powers our real-time 30+ language interface translation, cross-border contributor accessibility, and low-latency cloud model benchmarking pipelines.
Telegram Open Network
Zero-Knowledge Communication Enabler
Enables end-to-end encrypted secret chats for host credential handover, real-time admin dispatch, and 24/7 technical customer support.
Cloudflare
Session Isolation & Security Enabler
Secures intake endpoints against malicious bots and wraps remote desktop tasker sessions in strict zero-trust perimeter policies.
24/7 Response Desk
Contact Us & Direct Leadership Directory
Send an Encrypted Inquiry
Connect immediately with verified customer care agents and leadership admins to speed up account audits, review qualifications, or request accommodations.
Direct Team Channels
- Marcus Vance · Managing PartnerPartnerships & AccountsEmail →
- Dr. Elena Rostova · Learning LeadTasker Curriculum & OnboardingEmail →
- Aris Thorne · Head of QASampling Audits & ScorecardsEmail →
- Kaelen Diaz · Security LeadEncrypted Channels & VM SessionsEmail →
- Sora Tanaka · Contributor CareAccessibility & 24/7 TG SupportEmail →
- Legal & Compliance:
- legal@middlemen.ai
- Security Vulnerabilities:
- security@middlemen.ai
- Governing Entity:
- Middlemen (MM) Ltd.