Story S.

Key Skills

Software

Data Annotation Tech

Internal/Proprietary Tooling

Top Subject Matter

1. Finance – Accounting & Financial Reporting

2. Finance – Risk Analysis & Fraud Detection

3. Business / Professional Services – Document Review & Analysis

Top Data Types

Text

Document

Top Task Types

RLHF

Evaluation/Rating

Prompt + Response Writing (SFT)

Question Answering

Function Calling

Text Summarization

Text Generation

Freelancer Overview

I am an accounting and audit professional with nearly four years of experience across KPMG and FGMK, combined with current contract work as an AI Training Contributor evaluating large language model (LLM) responses in accounting and finance domains. My background includes financial statement audit, journal entry logic, revenue recognition, internal controls, and US GAAP research, which allows me to assess model outputs for technical accuracy, reasoning quality, and domain reliability in high-stakes business use cases. In my AI training work, I evaluate LLM and agent responses, write expert prompts and gold-standard reference answers, perform RLHF-style preference ranking, and document recurring failure modes such as factual inaccuracies, unsupported reasoning, and calculation or reconciliation errors. I also bring Python, SQL, Excel, and audit analytics experience, which helps me work effectively with structured data, technical workflows, and detail-oriented quality review.

Labeling Experience

RLHF Preference Ranking - Accounting and Finance Domain

DocumentRLHF

Performed RLHF-style preference ranking on large language model (LLM) responses in accounting, audit, tax, journal-entry, and financial-analysis tasks. Compared competing model outputs and selected stronger responses based on factual accuracy, reasoning quality, completeness, and alignment with accounting and finance domain standards, including US GAAP concepts such as ASC 606, ASC 842, ASC 740, consolidations, and journal entry logic. Created expert prompts and gold-standard reference responses to support ranking and training workflows, and documented recurring model failure patterns such as unsupported conclusions, factual inaccuracies, weak calculations, and domain-specific accounting errors. This work helped improve model quality for high-stakes finance and accounting use cases.

2026 - Present

LLM Evaluation & AI Training - Accounting and Finance Domain

DocumentEvaluation/Rating

Evaluated LLM and agent responses on accounting, audit, tax, journal-entry, and financial-analysis prompts, scoring outputs for factual accuracy, reasoning quality, and tool-use behavior. Created expert prompts and gold-standard reference responses, performed RLHF-style preference ranking, fact-checked AI-generated accounting content, and documented recurring error patterns to improve model quality.

2026 - Present

Education

Work History