Constantine S.

Key Skills

Software

AWS SageMaker

Appen

Top Subject Matter

AI Text Humanization

Legal Automation

Legal Services & Contract Review

Top Data Types

Text

Document

Top Task Types

Fine-tuning

Freelancer Overview

CONSTANTINE STANLEY +91 9895942514 | [email protected] linkedin.com/in/constantine-stanley-aab24424 Trivandrum, Kerala 695011, India PROFESSIONAL SUMMARY AI Data Labeling Specialist with 3+ years of experience building and scaling annotation pipelines for large language models. Expert in human-in-the-loop (HITL) workflows, LLM-assisted pre-labeling, and quality assurance. Proven ability to curate high-accuracy labeled datasets across legal, government, and retail domains, reducing manual labeling effort by up to 70% while maintaining 95%+ accuracy. CORE SKILLS • AI-Assisted Annotation & Pre-Labeling: Claude, GPT-4o, Mistral 7B, DeepSeek-R1, Gemini via AWS Bedrock & GCP Vertex AI • Annotation & HITL Platforms: AWS SageMaker Ground Truth, custom UIs, active learning • Data Quality & Validation: Inter-annotator agreement, defect tracking, Six Sigma gates • Data Processing: Python, SQL, LangChain, HuggingFace, tokenization, chunking, NER • Pipeline & Storage: Apache Spark, Airflow, Snowflake, Cassandra, PostgreSQL • Cloud & DevOps: AWS (EC2, S3, Lambda, EKS), Docker, Kubernetes, Git, CI/CD PROFESSIONAL EXPERIENCE AI Data Labeling Specialist May 2021 – Present (3+ Years) • Designed end-to-end annotation workflows for 10,000+ legal documents per month, integrating LLM pre-labeling (Claude, GPT-4o, DeepSeek-R1) to automate initial tagging and cut manual review time by 60%. • Implemented human-in-the-loop (HITL) systems with active learning, prioritizing uncertain samples for human annotators; reduced labeling costs by 40% while sustaining 95%+ inter-annotator agreement. • Curated and managed labeled datasets for Retrieval-Augmented Generation (RAG) pipelines across multiple jurisdictions, supporting real-time document understanding. • Built automated data preprocessing pipelines (Kafka → Spark → Snowflake) to clean and prepare raw text for annotation, slashing data latency from hours to <90 seconds. • Configured and operated annotation projects on AWS SageMaker Ground Truth and custom web-based labeling interfaces, leveraging spot instances for cost efficiency. • Enforced Six Sigma-based quality gates and defect tracking, reducing labeling errors by 30% through iterative feedback loops. • Collaborated with ML/NLP engineers to refine annotation guidelines, tokenization strategies, and model fine-tuning datasets. EDUCATION MCA (Master of Computer Applications) Manomaniam Sundaranar University

Labeling Experience

Ai data labeling specialist

ImageClassification

Image classification

2021 - 2026

Education

Work History