Oksana S.

Key Skills

Software

Other

Top Subject Matter

LLM evaluation

human feedback

model response ranking

Top Data Types

Video

Top Task Types

Function Calling

Classification

Freelancer Overview

LLM Evaluation Platform - TypeScript / AI evaluation / human feedback. Brings 14+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Other and Hugging Face. Education includes Bachelor of Arts (Honours), London Metropolitan University (2016). AI-training focus includes data types such as Computer Code and Programming and labeling workflows including Evaluation, Rating, and Function Calling.

Labeling Experience

PayLabel AI - Fintech data-labelling

OtherClassification

Developed fintech payment-event labeling for operational and risk-related categories. Labeled instances covering failed payments, chargebacks, fraud signals, disputes, and KYC/AML-style operational tags. Produced structured labels suitable for downstream model training and operational analytics. • Failed payment and chargeback event labeling • Fraud signal and dispute classification • KYC/AML-style operational label definitions • Labeled dataset creation for fintech use cases

2020 - Present

Data Annotation Tool + Hugging Face - TypeScript / AI training data

Function Calling

Created an annotation workflow tool for categories, tags, sentiment, intent, and classification labels. Ran model-assisted labeling experiments using Hugging Face model usage. Focused on transforming raw examples into structured training/benchmark datasets using repeatable labeling procedures. • Taxonomy design for categories/tags/sentiment/intent • Annotation workflows and label assignment consistency • Model-assisted labelling experiments with Hugging Face • Preparation of labeled datasets for AI training

2020 - Present

LLM Evaluation Platform - TypeScript / AI evaluation / human feedback

Other

Built an AI response review system to compare model outputs and apply human-feedback-informed scoring. Implemented evaluation criteria covering answer quality, accuracy, and usefulness. Used rubric-driven judgments to support model output comparison and iterative improvement. • Response comparison and scoring logic • Accuracy/usefulness evaluation criteria • Human feedback decision workflows • Dataset/evaluation preparation for further experiments

2020 - Present

Education

Work History