Revathi Selvaraj

Key Skills

Software

Appen

Labelbox

Scale AI

Surge AI

Remotasks

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Document

Text

Top Label Types

Entity Ner Classification

Classification

Text Generation

Emotion Recognition

Freelancer Overview

I am an AI FinTech Developer and Quantitative Analyst with hands-on experience in building and optimizing machine learning pipelines, including data labeling, annotation, and quality control for AI training data. My background includes developing and fine-tuning LLM-driven modules using NLP techniques, as well as implementing supervised and unsupervised machine learning models for time-series forecasting and analytics. I have completed a Data Labeling Job Simulation (Forage), where I gained practical skills in batch labeling, PII awareness, and iterative quality control, and have applied these skills in real-world projects such as inventory intelligence modules and procurement analytics dashboards. Proficient in Python, Pandas, NumPy, and Power BI, I am adept at designing data workflows, ensuring data quality, and collaborating with cross-functional teams to deliver accurate and high-value AI solutions. My experience spans finance, supply chain, and enterprise analytics domains, with a strong focus on data-driven decision-making and stakeholder communication.

Labeling Experience

Python Full Stack Developer Intern (Inmakes Infotech)

Scale AITextEntity Ner ClassificationClassification

Structured and validated API response datasets for frontend consumption, applying schema labeling and data-type annotation to ensure consistency across PostgreSQL pipelines. Annotated UI component behavior and edge-case states to support model-driven testing workflows. Developed RESTful APIs using Django serializers consumed by React.js frontend components, optimizing data flow and reducing response time by 35%

Structured and validated API response datasets for frontend consumption, applying schema labeling and data-type annotation to ensure consistency across PostgreSQL pipelines. Annotated UI component behavior and edge-case states to support model-driven testing workflows. Developed RESTful APIs using Django serializers consumed by React.js frontend components, optimizing data flow and reducing response time by 35%

2025

Financial Derivatives Data Labeler (Skew Sculptor Studio - ZeTheta Algorithms)

Scale AIDocumentEntity Ner ClassificationClassification

While working on Skew Sculptor Studio’s Exotic Derivatives Pricing Engine (ZeTheta Algorithms), I labeled over 10,000 Monte Carlo simulation paths with payoff classifications. The role required application of bump-and-revalue methods to produce sensitivity labels for model validation. Annotated ground-truth data formed the basis for evaluating pricing and sensitivity models. • Labeled simulation data with correct payoff class (Asian, Lookback, Barrier). • Generated delta/vega sensitivity labels with quantitative methods. • Supported supervised ML pricing model performance evaluation. • Ensured dataset compliance with ground-truth requirements.

While working on Skew Sculptor Studio’s Exotic Derivatives Pricing Engine (ZeTheta Algorithms), I labeled over 10,000 Monte Carlo simulation paths with payoff classifications. The role required application of bump-and-revalue methods to produce sensitivity labels for model validation. Annotated ground-truth data formed the basis for evaluating pricing and sensitivity models. • Labeled simulation data with correct payoff class (Asian, Lookback, Barrier). • Generated delta/vega sensitivity labels with quantitative methods. • Supported supervised ML pricing model performance evaluation. • Ensured dataset compliance with ground-truth requirements.

2025 - 2025

Financial Data Labeler (Greeks Gym - ZeTheta Algorithms)

Scale AIDocumentEntity Ner ClassificationClassification

On the Greeks Gym Options Analytics Platform project with ZeTheta Algorithms, I generated structured financial labels for simulated financial instrument scenarios. My work included producing annotated datasets that linked input parameters (strike price, volatility) to model payoff outcomes. These annotated datasets supported subsequent model training tasks for option analytics. • Created labels for simulated scenarios including Greeks (Delta, Gamma, etc.). • Mapped combinations of option parameters to training outputs. • Produced annotation-ready payoff datasets for ML models. • Used Python and relevant analytics libraries for labeling.

On the Greeks Gym Options Analytics Platform project with ZeTheta Algorithms, I generated structured financial labels for simulated financial instrument scenarios. My work included producing annotated datasets that linked input parameters (strike price, volatility) to model payoff outcomes. These annotated datasets supported subsequent model training tasks for option analytics. • Created labels for simulated scenarios including Greeks (Delta, Gamma, etc.). • Mapped combinations of option parameters to training outputs. • Produced annotation-ready payoff datasets for ML models. • Used Python and relevant analytics libraries for labeling.

2025 - 2025

Data Labeling Job Simulation - Data Labeling Specialist (Forage)

Scale AITextEntity Ner ClassificationClassification

Completed an end-to-end data labeling workflow via Forage's Data Labeling Job Simulation. Tasks included batch labeling of unstructured text, PII detection and redaction (names, emails, phone numbers, financial identifiers), iterative quality control, inter-annotator agreement review, and annotation guideline improvement. Achieved structured label outputs with zero PII false-negatives.

Completed an end-to-end data labeling workflow via Forage's Data Labeling Job Simulation. Tasks included batch labeling of unstructured text, PII detection and redaction (names, emails, phone numbers, financial identifiers), iterative quality control, inter-annotator agreement review, and annotation guideline improvement. Achieved structured label outputs with zero PII false-negatives.

2025 - 2025

NLP Intent Classification & LLM Label Creation

TextClassification

Annotated 10,000+ NLP training records for intent classification and entity span validation across enterprise inventory query datasets. Created ground-truth Q&A answer labels for GPT-based LLM fine-tuning on domain-specific data. Evaluated LLM model outputs against human-annotated benchmarks achieving 95% query accuracy. Maintained version-controlled annotation logs and ensured label consistency across all batch submissions. Reduced manual data preparation time by 30% via automated labeling pipelines with human-in-the-loop validation checkpoints.

Annotated 10,000+ NLP training records for intent classification and entity span validation across enterprise inventory query datasets. Created ground-truth Q&A answer labels for GPT-based LLM fine-tuning on domain-specific data. Evaluated LLM model outputs against human-annotated benchmarks achieving 95% query accuracy. Maintained version-controlled annotation logs and ensured label consistency across all batch submissions. Reduced manual data preparation time by 30% via automated labeling pipelines with human-in-the-loop validation checkpoints.

2023 - 2024

Show more

Education

Work History