Chika O.

Key Skills

Software

No software listed

Top Subject Matter

Legal - lease agreement + vendor contracts

Artificial intelligence -

Finance - Vendor agreement, payment terms, invoices

Top Data Types

Text

Document

Top Task Types

Classification

Entity (NER) Classification

Evaluation/Rating

Text Summarization

Text Generation

Data Collection

Freelancer Overview

IT Officer (Technical Support & Operations) | Tchkyas Marine Services | 2018 - Present 6+ years professional experience in high-accuracy data workflows, QA processes, and technical documentation. AI Training & Data Experience: - Structured data extraction: Converted PDFs and contracts into validated JSON following strict schemas - Text annotation: Classification, entity labeling, and quality rating for NLP datasets - Data types: Text, legal documents, vendor agreements, financial records - Labeling workflows: Schema validation, null-field handling, date/currency normalization, table-to-array conversion - Tools: JSONLint, Excel, Google Sheets, annotation platforms - Achieved 100% syntax accuracy on complex nested objects and 3-level data structures Education: B.Sc. Computer Science, Crawford University (2009) Available 25+ hrs/week. Nigeria-based with fiber internet and dedicated workstation.

Labeling Experience

LLM Output Evaluation & JSON Quality Rating

TextEvaluation/Rating

Independent AI training focused on LLM output evaluation and error correction for structured data tasks. Scope: Reviewed AI-generated JSON drafts against source legal documents and strict schemas. Rated outputs for accuracy, schema compliance, and data integrity. Key evaluation tasks: - Compared AI JSON to source PDFs: identified missing entities, wrong dates, type errors - Rated outputs on criteria: field accuracy, null-handling, case sensitivity, number formatting - Identified common LLM errors: DD/MM vs MM/DD dates, string/number type mismatches, hallucinated fields - Rewrote incorrect AI outputs to produce gold-standard JSON for SFT datasets - Provided error categories and correction guidelines for model improvement Quality: Applied 15-point validation checklist covering syntax, schema, and factual accuracy. Tools: JSONLint, diff checkers, schema validators, annotation guidelines.

2026 - Present

Text Summarization & Key Fact Extraction for Legal Contracts

TextText Summarization

Independent AI training focused on abstractive summarization for LLM datasets. Scope: Created original summaries for complex legal/financial documents including lease agreements, vendor contracts, and service agreements. All summaries written from extracted entities without copy-paste. Key tasks: - Identified critical facts: parties, dates, financial terms, obligations, termination clauses - Wrote concise summaries per guidelines: 1-3 sentences, present tense, no opinions - Ensured 100% fact coverage: effective dates, payment terms, contract values, key contacts - Rewrote AI-generated draft summaries to remove hallucinations and source-mismatches - Validated summaries against source PDFs for accuracy and completeness Quality: Maintained factual accuracy and guideline compliance across multi-page documents. Tools: Source document review, style guides, fact-checking checklists.

2026 - Present

Legal Document Entity Extraction & JSON Schema Annotation

TextEntity (NER) Classification

Independent AI training project for LLM datasets focused on Named Entity Recognition and structured data extraction. Scope: Converted complex legal/financial PDFs including lease agreements, vendor contracts, and SOWs into validated JSON following strict schemas. Key NER/annotation tasks: - Identified and extracted key entities: parties, dates, monetary values, addresses, RC numbers, contact details - Normalized entities: dates to YYYY-MM-DD, currencies, numbers, text case per schema - Structured nested entities and arrays from tables: services, representatives, contact objects - Handled null-entity cases for missing required schema fields - Performed schema validation to ensure correct entity types: string vs number compliance - Wrote original summaries capturing all extracted entities, no copy-paste Quality: Achieved 100% accuracy on 3-tier test cases with 3-level entity nesting and multi-format dates. Tools: JSONLint, Excel, Google Sheets, schema documentation.

2026 - Present

Education

Work History