Ozioma Isaiah

Full-Stack AI Engineer and LLM Specialist with 6+ years of experience designing, building, and deploying production-grade web applications and AI-powered systems. Experienced in owning full-stack architectures (Django, FastAPI, React), integrating generative AI (LLMs, RAG pipelines, vector databases), and building machine learning workflows from data preprocessing through model evaluation. Proven contributor to RLHF training pipelines, evaluating response quality, identifying failure modes, authoring high-quality preference data, and designing domain-specific evaluation rubrics.

Equally comfortable writing Python microservices, React frontends, or fine-tuning prompt strategies for production AI systems. Fluent communicator who translates complex technical concepts clearly for both technical and non-technical audiences.

AI Code Evaluator / LLM Trainer at Alignerr (Labelbox) (2025 – Present)

• Evaluated and compared AI-generated code responses across multiple frontier models, focusing on logical correctness, edge case handling, and alignment with prompt requirements.
• Applied structured evaluation frameworks assessing outputs across logic, naming clarity, modularity, interface design, error handling, documentation, and overall review readiness.
• Performed detailed code reviews identifying bugs, inefficient implementations, and inconsistencies, ensuring solutions were production-ready and aligned with best engineering practices.
• Conducted response ranking and comparative analysis (Model A vs Model B), providing clear, evidence-based justifications for preference decisions.
• Analyzed model behaviour to detect hallucinations, weak reasoning, and incomplete implementations, contributing to improved model reliability.
• Reviewed and validated API designs, function signatures, and system interactions for usability, maintainability, and adherence to standard design patterns.
• Contributed to training data quality by refining evaluation criteria, improving prompt clarity, and ensuring high-quality feedback loops for model improvement workflows.

AI Trainer & LLM Response Evaluator at Outlier AI (Scale AI) (2024 – Present)

• Evaluated 1,000+ LLM-generated responses per month for accuracy, helpfulness, coherence, and safety, producing preference rankings used directly in RLHF training pipelines.
• Authored high-quality model responses across STEM, coding, reasoning, and open-ended generation tasks, maintaining quality scores consistently above platform benchmarks.
• Developed evaluation rubrics and scoring guidelines for novel task categories, accelerating onboarding for new raters on the team.
• Identified and flagged systematic failure patterns in model outputs — including hallucinations, instruction non-adherence, and harmful completions — submitting detailed bug reports that informed model iteration.
• Completed specialised coding evaluation tracks in Python and JavaScript, assessing code correctness, efficiency, readability, and security.
• Maintained 98%+ task acceptance rate across all project categories over a 12-month period.

Search Engine Evaluator at TELUS International (2022 – 2023)

• Analysed and rated search engine results for relevance and intent alignment across diverse query categories.
• Tagged and structured datasets to support machine learning model improvement for search quality.
• Delivered structured quality reports to global teams, maintaining consistent evaluation standards.

AI / ML / Full-Stack Engineer at Freelance / Contract (2017 – Present)

• Delivered 20+ production-ready web applications integrating Django/FastAPI backends with React frontends across fintech, edtech, and e-commerce verticals.
• Architected scalable REST APIs and authentication systems (JWT, role-based access control) for multi-tenant platforms.
• Built AI-powered systems leveraging LLMs, RAG pipelines, and vector databases (FAISS, Pinecone) for context-aware generative AI applications.
• Designed interactive HTML5 Canvas-based UI systems, business dashboards, and data visualisation components.
• Implemented QA testing pipelines (unit, regression, API testing) using Selenium, Postman, and Jenkins to prevent regressions.
• Deployed applications using modern cloud platforms (Render, Vercel, AWS); containerised systems with Docker and managed CI/CD via GitHub Actions.
• Mentored junior developers, conducting code reviews and pair-programming sessions on React and Python best practices.

QA Tester Intern at LinuxJobber (2019 – 2020)

• Designed and executed functional and regression test cases for Django and React web applications.
• Automated test scripts using Selenium, reducing manual testing effort by approximately 40%.
• Conducted API testing using Postman and integrated CI pipelines with Jenkins.
• Supported QA processes across multiple software releases, improving defect detection rates.

B.Sc. in Statistics – Federal University of Technology, Nigeria (2016 – 2022)