For employers

Hire this AI Trainer

Sign in or create an account to invite AI Trainers to your job.

LoginCreate account
Invite to Job
N
Nguyen Quynh

Nguyen Quynh

Agency
Vietnam flagHanoi, Vietnam
1bitlab.io/
$16.00/hrExpert40+

Key Skills

Software

LabelboxLabelbox
LabelImgLabelImg

Top Subject Matter

No subject matter listed

Top Data Types

ImageImage
VideoVideo
Computer Code ProgrammingComputer Code Programming

Top Task Types

Object DetectionObject Detection
Fine-tuningFine-tuning
ClassificationClassification
Computer Programming/CodingComputer Programming/Coding
Text SummarizationText Summarization

Company Overview

AI Training Data Services — Built for Domain-Specific AI 1BITLAB is a Vietnam-based AI development company specializing in training data annotation and fine-tuning across four verticals: accounting & finance, Web3/blockchain, fintech, and education. Our annotators aren't generalist crowdworkers. They're paired with engineers who've built production systems in these domains — which means labels that reflect real-world context, not just surface-level pattern matching. Why teams choose us: Domain expertise across financial documents, DeFi protocols, e-learning content, and fintech workflows Engineering-backed QA — multi-stage review with guideline calibration, inter-annotator checks, and final audit before every delivery Mid-market pricing — built for seed-to-Series B AI teams, without enterprise overhead Language coverage — English, Japanese, Korean, Vietnamese We handle RLHF preference labelling, document classification, intent annotation, fine-tuning datasets, OCR correction, and multilingual NLP — delivered in your format, integrated with your pipeline. Fast turnaround. Senior team. No layers between you and the work.

ExpertEnglishJapaneseVietnameseKorean

Security

Security Overview

1BITLAB maintains strict data security and privacy practices aligned with ISO 27001 standards across all annotation projects. Data Handling: - All client data is processed under signed NDA and data processing agreements - Project data is stored in isolated environments — no cross-project data access - Data is deleted or returned to client upon project completion, per agreement terms Access Control: - Role-based access: annotators access only their assigned task batches - No data downloaded to personal devices - VPN and secure login required for all remote access Physical Security: - Dedicated secure room available for sensitive projects requiring on-site, air-gapped annotation - Access restricted to assigned team members only during project duration Annotator Agreements: - All team members sign confidentiality agreements before project onboarding - Domain-specific NDAs available upon client request We are happy to accommodate additional security requirements on a per-project basis.

Labeling Experience

Accounting intention

OtherTextClassification
Ongoing project supporting the development of an AI-powered accounting data integration platform that automatically consolidates financial records from multiple sources into SAP. Tasks performed: - Labeling and classification of accounting transactions by category: revenue, expense, asset, liability, tax entries - Entity extraction annotation from raw financial documents: vendor name, invoice number, amount, currency, date, cost center - Mapping annotation: tagging source-field to SAP chart-of-accounts (CoA) equivalents to train the auto-mapping model - Data quality flagging: identifying inconsistent formats, duplicate entries, and ambiguous classifications across source systems - Validation labeling for edge cases: multi-currency transactions, intercompany entries, partial payments Data sources covered: ERP exports, CSV/Excel uploads, bank statements, PDF invoices

Ongoing project supporting the development of an AI-powered accounting data integration platform that automatically consolidates financial records from multiple sources into SAP. Tasks performed: - Labeling and classification of accounting transactions by category: revenue, expense, asset, liability, tax entries - Entity extraction annotation from raw financial documents: vendor name, invoice number, amount, currency, date, cost center - Mapping annotation: tagging source-field to SAP chart-of-accounts (CoA) equivalents to train the auto-mapping model - Data quality flagging: identifying inconsistent formats, duplicate entries, and ambiguous classifications across source systems - Validation labeling for edge cases: multi-currency transactions, intercompany entries, partial payments Data sources covered: ERP exports, CSV/Excel uploads, bank statements, PDF invoices

2026 - Present

Validate AI model

OtherComputer Code ProgrammingFine TuningComputer Programming Coding
Ongoing project focused on human verification and quality evaluation of AI-generated source code across multiple programming languages, including Python and PHP. Tasks performed: - Line-by-line review of AI-generated code for correctness, logic errors, and security vulnerabilities - Functional correctness labeling: pass / fail / needs revision per code snippet - Preference ranking between multiple AI-generated solutions for the same prompt (RLHF-style) - Error categorization: syntax error, logic flaw, security risk, style violation, hallucinated API - Annotation of code comments and docstrings for clarity and accuracy - Edge case flagging for ambiguous or unsafe outputs Languages covered: Python, PHP (expandable per client requirement) Quality measures: - All reviewers are practising software engineers, not general annotators - Dual-review on flagged or borderline snippets - Weekly calibration sessions to align judgment criteria across the team

Ongoing project focused on human verification and quality evaluation of AI-generated source code across multiple programming languages, including Python and PHP. Tasks performed: - Line-by-line review of AI-generated code for correctness, logic errors, and security vulnerabilities - Functional correctness labeling: pass / fail / needs revision per code snippet - Preference ranking between multiple AI-generated solutions for the same prompt (RLHF-style) - Error categorization: syntax error, logic flaw, security risk, style violation, hallucinated API - Annotation of code comments and docstrings for clarity and accuracy - Edge case flagging for ambiguous or unsafe outputs Languages covered: Python, PHP (expandable per client requirement) Quality measures: - All reviewers are practising software engineers, not general annotators - Dual-review on flagged or borderline snippets - Weekly calibration sessions to align judgment criteria across the team

2024 - Present
Labelbox

Beverage recognition

LabelboxImageObject DetectionClassification
Annotated image and video datasets to train a robotic arm system for automated beverage recognition and handling in supermarket environments. Tasks performed: - Bounding box annotation of beverage products across multiple SKUs (bottles, cans, cartons, pouches) - Object classification by product type, brand, and packaging format - Keypoint labeling for grip-point detection to guide robotic arm positioning - Occlusion and edge-case flagging for partially visible or stacked products - Multi-angle image annotation to support 3D spatial recognition Quality measures: - Annotation guidelines calibrated with client's robotics engineering team - Two-stage review: peer check + senior QA audit per batch - Consistency checks across similar SKUs to reduce misclassification

Annotated image and video datasets to train a robotic arm system for automated beverage recognition and handling in supermarket environments. Tasks performed: - Bounding box annotation of beverage products across multiple SKUs (bottles, cans, cartons, pouches) - Object classification by product type, brand, and packaging format - Keypoint labeling for grip-point detection to guide robotic arm positioning - Occlusion and edge-case flagging for partially visible or stacked products - Multi-angle image annotation to support 3D spatial recognition Quality measures: - Annotation guidelines calibrated with client's robotics engineering team - Two-stage review: peer check + senior QA audit per batch - Consistency checks across similar SKUs to reduce misclassification

2020 - 2020
Labelbox

Online studying evaluation

LabelboxVideoClassificationObject Detection
Developed a video-based student behavior annotation pipeline for an online education platform. The project involved labeling captured classroom session recordings to evaluate student engagement, attention levels, and learning activity patterns. Tasks performed: - Frame-level annotation of student facial expressions and body posture using Labelbox - Classification of engagement states: focused, distracted, absent, active participation - Temporal segmentation of learning activity sequences - Edge case review and label reconciliation across annotator team Project size: ~15,000 video segments annotated across multiple cohorts.

Developed a video-based student behavior annotation pipeline for an online education platform. The project involved labeling captured classroom session recordings to evaluate student engagement, attention levels, and learning activity patterns. Tasks performed: - Frame-level annotation of student facial expressions and body posture using Labelbox - Classification of engagement states: focused, distracted, absent, active participation - Temporal segmentation of learning activity sequences - Edge case review and label reconciliation across annotator team Project size: ~15,000 video segments annotated across multiple cohorts.

2019 - 2020