Somtochukwu A.

Key Skills

Software

Don't disclose

Top Subject Matter

LLM evaluation

Qa Domain Expertise

and rubric-based human feedback for AI responses

Top Data Types

Text

Document

Image

Top Task Types

Tracking

RLHF

Entity (NER) Classification

Freelancer Overview

Self-directed practice in AI output evaluation, prompt assessment, and QA (rubric-based rating and error taxonomy). Brings 3+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Don't disclose, OpenCV, and Google Docs. Education includes Bachelor of Engineering, Federal University of Technology Owerri (2024) and West African Senior School Certificate, Federal Science and Technical College, Yaba (2024). AI-training focus includes data types such as Text, Document, and Image and labeling workflows including Evaluation, Rating, and

Labeling Experience

Innovation Challenge Participant - Julius Berger Nigeria

ImageTrackingRLHF

You participate in a technology challenge by producing written specifications for an AI-driven computer vision pipeline. You define model input/output requirements, labelling criteria, and evaluation standards to guide consistent downstream work. You also create structured documentation for dataset requirements and quality benchmarks using professional data documentation rigor. • Develop written specifications for AI model I/O requirements • Define labelling criteria and evaluation standards for a computer vision pipeline • Produce documentation for dataset requirements, annotation guidelines, and quality benchmarks • Apply structured, benchmark-driven documentation practices to support evaluation workflows

2025 - Present

Innovation Challenge participant: MATQUAL Scanner documentation of labeling and evaluation criteria

Text

Developed written specifications that define AI model input/output requirements, dataset requirements, labeling criteria, and evaluation standards for a computer vision pipeline. Produced documentation describing annotation guidelines and quality benchmarks to support consistent data labeling and downstream model evaluation. Applied professional data-labeling rigor by translating technical requirements into structured, reviewable instructions suitable for QA and alignment. • Defined labeling criteria and evaluation standards for a TensorFlow Lite / OpenCV vision pipeline • Documented dataset requirements and annotation guidelines • Established quality benchmarks for labeled data outputs • Created structured, guideline-driven documentation for consistent labeling and evaluation

2025 - Present

Self-directed practice in AI output evaluation, prompt assessment, and QA (rubric-based rating and error taxonomy)

Don't discloseText

Conducted rubric-based evaluation of AI-generated responses using a personal error taxonomy to categorize failures such as hallucination, omission, tone mismatch, formatting errors, and ambiguous reasoning. Performed prompt/response grading against defined quality criteria including clarity, instruction adherence, logical consistency, and appropriate tone, while tracking safety, bias, and policy compliance. Applied structured quality review practices consistent with human feedback and AI training workflows for independent model evaluation. • Rubric design and rating across multiple response dimensions (accuracy, logic, clarity, tone, instruction adherence) • Error taxonomy development for repeatable analysis • Safety, bias, toxicity, and guideline-violation flagging using publicly available AI systems • Structured QA review workflow for consistent outputs

2024 - Present

Technical contributor: Enactus FUTO sustainability project (structured research synthesis and accuracy checks)

Document

Synthesised research into structured reports while applying consistency checks to ensure information accuracy across multiple deliverables. Conducted review-style verification to align written outputs with source material and intended findings, mirroring structured QA approaches used in annotation and evaluation workflows. Produced documentation summarizing methodology and recommendations with an emphasis on clarity and correctness. • Cross-source research synthesis into structured outputs • Consistency checks to validate accuracy across deliverables • QA-oriented documentation review before finalising reports • Clear structuring of methodology and recommendations

2024 - 2024

Education

Work History