Nehemiah Anyidakuzo Shekwoyeyilo

Algorithmic Instruction Tuning & CoT Reasoning: Designed adversarial prompts and multi-turn debugging scenarios to test and refine LLM Chain-of-Thought reasoning across complex Python data structures and algorithms. Code Auditing & Security Reviews: Executed comparative reviews of AI-generated Python and web framework code against strict PEP 8 compliance, checking for runtime errors, logic fallacies, and security risks (e. g., SQL injection, memory leaks). ML Library & Data Engineering for SFT: Developed Python scripts to extract, clean, structure, and validate datasets for Supervised Fine-Tuning using Pandas, NumPy, and Scikit-learn. Full-Stack Development (Pre Mark): Architected and deployed Pre Mark, a full-stack data annotation application built with a FastAPI (Python) backend, SQLite persistence layer, and React frontend. Web Scraping & Data Extraction Pipelines: Engineered end-to-end web scraping workflows using Selenium, Beautiful Soup, and Requests to extract, clean, and output structured JSON/CSV data for NLP fine-tuning tasks. Annotation & QA Leadership: Managed a multi-annotator data pipeline: defined specifications, set acceptance criteria, conducted QA code audits, and improved task feedback loops.

Built custom Python scripts to automate stock tracking, process operational data, and deliver structured reporting, reducing manual data errors and achieving 100% audit accuracy. Wrote analytical technical documentation and performance reports to establish measurable KPIs across workflow pipelines.

Designed structured, rubric-based educational frameworks from first principles — transferable to constructing well-scoped Python agent evaluation benchmarks.

Managed remote field team compliance and daily structured reporting chains in an outcome-focused environment.