Adeboye B.

Key Skills

Software

Other

Data Annotation Tech

Top Subject Matter

Software code defect detection

AI-assisted code review

hybrid ML-human review framework

Top Data Types

Text

Image

Video

Top Task Types

Text Generation

RLHF

Transcription

Data Collection

Prompt + Response Writing (SFT)

Freelancer Overview

AI/ML Research — Hybrid Random Forest-Human Sequential Code Review Framework (MSc dissertation). Brings 6+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Internal, Proprietary Tooling, and Other. Education includes Master of Science, Nile University of Nigeria (2026) and Bachelor of Science, Bingham University (2023). AI-training focus includes data types such as Computer Code and Programming and labeling workflows including Computer Programming, Coding, and Text Generation.

Labeling Experience

AI Engineer — CodeSentry: AI Code Review Intelligence Platform (In Development)

RLHF

Building a deployable developer tool for AI-assisted code review risk scoring that routes PRs between AI and human review. Implements an ML pipeline that mines GitHub PR metadata, performs feature engineering, and generates real-time defect risk scores with confidence thresholds. Augments review outputs with LLM-generated summaries using structured prompt templates and enforced JSON output schemas. • GitHub API data mining and PR feature engineering for risk scoring • Random Forest classifier-based risk scoring and decision routing (low-risk AI review vs high-risk human review) • Next.js/TypeScript + PostgreSQL/Prisma application with Python scikit-learn microservice • LLM-augmented structured review summaries with schema enforcement

2026 - Present

AI Engineer — VaultAI: RAG-Powered Document Intelligence Platform (In Development)

OtherText Generation

Developing an AI knowledge-base product using Retrieval-Augmented Generation (RAG) with grounded, source-cited answers. Designs retrieval and prompting strategies including hybrid search, re-ranking, and query decomposition to improve answer correctness across heterogeneous documents. Implements strict prompt guardrails and citation enforcement to reduce hallucinations in generated responses. • Document ingestion and chunking strategy for PDF/DOCX/TXT and vector embedding via OpenAI • Semantic retrieval using pgvector and PostgreSQL plus hybrid semantic-keyword BM25 retrieval • Conversational QA interface with grounded LLM responses constrained to retrieved context • Multi-tenant access controls, namespaces, and admin tooling for usage analytics and query history

2026 - Present

AI/ML Research — Hybrid Random Forest-Human Sequential Code Review Framework (MSc dissertation)

Conducted applied ML research to detect software code defects using an end-to-end pipeline built around GitHub pull-request mining and supervised model training. Implemented hybrid workflow elements to route AI-assisted reviews and compare outcomes against established baselines. Focused on extracting structured, LLM-generated review insights and enforcing output schemas for consistent evaluation. • Data sourcing from GitHub REST API across 50 open-source repositories and 723 pull requests • Feature engineering for PR risk scoring (e.g., PR size, complexity, contributor history, file churn) • Random Forest classifier training and 5-fold stratified cross-validation • Prompt engineering (few-shot, output schema enforcement, iterative prompt refinement) benchmarked vs CodeGuru, Copilot, and SonarCloud

2025 - 2026

Education

Work History