Oun A.

Key Skills

Software

Snorkel AI

Top Subject Matter

LLM alignment

RAG/AI agents

RLHF data preparation

Top Data Types

Text

Image

Document

Top Task Types

Function Calling

Data Collection

Object Detection

RLHF

Classification

Entity (NER) Classification

Segmentation

Bounding Box

Text Generation

Fine-tuning

Question Answering

Evaluation/Rating

Prompt + Response Writing (SFT)

Freelancer Overview

AI Engineer apprentice, MENADevs (AI agents & RAG; data annotation & LLM alignment). Brings 3+ years of professional experience across legal operations, contract review, compliance, and structured analysis. Core strengths include HuggingFace (T5, BART), and PyTorch. Education includes Bachelor's Degree, Al Hussein Technical University (HTU) (2026). AI-training focus includes data types such as Text and Image and labeling workflows including Function Calling, Data Collection, and Object Detection.

Labeling Experience

AI Engineer Apprentice - MENADevs

TextRLHF

Developed end-to-end AI agent and RAG-based chatbot solutions for multiple domain use cases, including tool calling, function execution, and contextual memory. Built evaluation and robustness tooling to support reliable agent behavior in industry scenarios, including database schema design and unit-test coverage. Applied LLM alignment concepts through workflow design while producing structured conversational data for downstream improvements. • Implemented RAG system chatbots and autonomous agents with real-time streaming and multi-modal processing • Authored testing utilities and unit tests for agent performance across scenario-based requirements • Designed ER database schemas and maintained evaluation/logging for failure analysis • Supported multi-turn conversation workflows with alignment-focused data preparation for RLHF and fine-tuning pipelines

2025 - 2026

AI Engineer apprentice, MENADevs (AI agents & RAG; data annotation & LLM alignment)

TextFunction Calling

Developed and evaluated multi-turn LLM workflows for RAG-based chatbots and AI agents across multiple domains, including conversation memory, context management, and real-time streaming. Authored testing artifacts and database schemas to ensure reliable agent behavior in industry scenarios such as banking systems. Performed LLM alignment data work by designing and comparing weak vs. strong model outputs using targeted nudges for RLHF and fine-tuning pipelines. • Designed multi-modal data processing for agent tool usage and real-time streaming • Engineered a bench-testing framework with unit tests and testing entries • Created RLHF/fine-tuning structured data via alignment nudges and workflow evaluation • Designed ER database schemas to support robust agent performance

2025 - 2026

AI & Machine Learning Trainee, Samsung Innovation Campus (YOLOv11 accident detection data engineering)

ImageObject Detection

Co-led data engineering for a YOLOv11-based accident detection system, including data preprocessing and augmentation. Performed training and evaluation using deep learning concepts and model metrics with hyperparameter tuning. Used common ML/DL toolchains to prepare models for real-world computer vision evaluation. • Prepared and augmented image datasets for YOLOv11 accident detection • Applied preprocessing steps to support training pipelines • Evaluated models with appropriate DL metrics and tuned hyperparameters • Trained/evaluated using Scikit-learn, TensorFlow, Keras, and PyTorch

2024 - 2025

AI Engineer Freelance, Jordan Innovation Campus (Multilingual AI agent; synthetic data & RAG integration)

TextData Collection

Built an end-to-end multilingual e-commerce conversational agent by generating and preparing domain-specific synthetic training data. Integrated a RAG pipeline optimized with LoRA/PEFT to improve responses using retrieval-augmented context. Orchestrated conversational flow with NLU tooling and implemented hybrid search to support semantic recommendations. • Generated synthetic domain-specific training data using T5/BART • Implemented RAG with LoRA (PEFT) optimization • Built hybrid search (TF-IDF + cosine similarity) and integrated RASA • Supported conversational flow management alongside LLM semantic recommendations

2024 - 2025

Education

Work History