For employers

Hire this AI Trainer

Sign in or create an account to invite AI Trainers to your job.

LoginCreate account
Invite to Job
E
Enoch Soyinka

Enoch Soyinka

Data Scientist & AI Training Specialist (Robotics, edgeAI, Computer Vision, NLP, RL & Bioinstrumentation)

Nigeria flagIle-Ife, Nigeria
$20.00/hrIntermediateLabelimgOpencv AI Kit OakRoboflow

Key Skills

Software

LabelImgLabelImg
OpenCV AI Kit (OAK)OpenCV AI Kit (OAK)
RoboflowRoboflow
DoccanoDoccano
Google Cloud Vertex AIGoogle Cloud Vertex AI

Top Subject Matter

Computer Vision & Robotics / Human-Computer Interaction
Natural Language Processing & Scientific Data
Bioinstrumentation & Signal Processing (Audio/Sensor Data)

Top Data Types

ImageImage
AudioAudio
TextText

Top Task Types

Object DetectionObject Detection
ClassificationClassification
Bounding BoxBounding Box
SegmentationSegmentation
Entity (NER) ClassificationEntity (NER) Classification
Point/Key PointPoint/Key Point
Fine-tuningFine-tuning

Freelancer Overview

I am a Data Scientist and Machine Learning Engineer with deep, hands-on experience building, labeling, and structuring custom datasets for complex AI models. In computer vision, I handle the entire data pipeline from collection to precise annotation. For example, I curated and labeled datasets to train a YOLOv5 model for a robotic arm that actively sorts apples, bananas, and oranges. For a Smart Vehicle Access Control System, I fine-tuned a model on a dataset of over 8,000 vehicles, handling bounding boxes for vehicle types and license plates, image segmentation for color recognition, and EasyOCR for text extraction. I also annotated and trained an autonomous vehicle perception system to reliably detect traffic lights, signs, and pedestrians at 92% mAP. Beyond standard image annotation, I work extensively with complex, multimodal data formats like audio spectrograms and skeletal landmarks. For a multi-factor authentication project, I processed raw voice audio into spectrogram images to train a ResNet50 model, pairing it with dlib for face recognition data. I have also generated structural training data using MediaPipe to extract exact hand and facial landmarks, feeding those coordinate datasets into custom TensorFlow models to control a robotic arm via hand gestures. Because I build the actual hardware and machine learning models, I know exactly how poor data labeling breaks a system in the real world, which is why I am meticulous about how training data is curated, validated, and fed into an AI pipeline.

IntermediateEnglishYorubaGerman

Labeling Experience

Autonomous Taxi Perception & Navigation Dataset on QLabs

ImageObject Detection
Collected 5000+ dataset (images and videos) within the QLabs(virtual environment) and annotated them to train a custom YOLOv5 perception module for an autonomous taxi. Classes includes traffic lights, street signs, and pedestrians.

Collected 5000+ dataset (images and videos) within the QLabs(virtual environment) and annotated them to train a custom YOLOv5 perception module for an autonomous taxi. Classes includes traffic lights, street signs, and pedestrians.

2026 - 2026

Materials Science NER & Parameter Extraction Dataset

TextEntity Ner Classification
Curated and manually annotated a highly specialized NLP dataset of 490 sentences sourced from scientific abstracts and research PDFs using the doccano annotation tool. Focused strictly on Named Entity Recognition (NER) to extract complex metal oxide synthesis parameters, including specific chemical precursors, temperature metrics, and gas sensitivity values. This meticulously labeled text data was then used to fine tune LLMs, specifically SciBERT and MatSciBERT, for automated data extraction.

Curated and manually annotated a highly specialized NLP dataset of 490 sentences sourced from scientific abstracts and research PDFs using the doccano annotation tool. Focused strictly on Named Entity Recognition (NER) to extract complex metal oxide synthesis parameters, including specific chemical precursors, temperature metrics, and gas sensitivity values. This meticulously labeled text data was then used to fine tune LLMs, specifically SciBERT and MatSciBERT, for automated data extraction.

2025 - 2025

Skeletal Hand Landmark Dataset for Gesture Control

VideoPoint Key Point
Handled data extraction and coordinate structuring using Intel RealSense camera on a QArm (Quanser Robotic Arm) and MediaPipe hand tracking. Extracted spatial point key points (hand landmarks) and formatted this coordinate data into a structured dataset to train a custom neural network capable of interpreting human gestures to control a QArm manipulator.

Handled data extraction and coordinate structuring using Intel RealSense camera on a QArm (Quanser Robotic Arm) and MediaPipe hand tracking. Extracted spatial point key points (hand landmarks) and formatted this coordinate data into a structured dataset to train a custom neural network capable of interpreting human gestures to control a QArm manipulator.

2025 - 2025

EEG Medical Data Annotator

OtherTextClassification
I curated and labeled clinical EEG time-series data to support supervised sleep stage classification modeling. Annotation work was carried out according to accepted scientific guidelines and contributed to research publication quality. Output supported both sleep staging and downstream LLM-enabled analysis platforms. • Annotated multiple EEG sessions using predefined classification categories. • Ensured rigorous cross-validation of assigned labels with domain experts. • Managed raw and processed data organization for research reproducibility. • Utilized YASA and internal tools for consistent annotation workflow.

I curated and labeled clinical EEG time-series data to support supervised sleep stage classification modeling. Annotation work was carried out according to accepted scientific guidelines and contributed to research publication quality. Output supported both sleep staging and downstream LLM-enabled analysis platforms. • Annotated multiple EEG sessions using predefined classification categories. • Ensured rigorous cross-validation of assigned labels with domain experts. • Managed raw and processed data organization for research reproducibility. • Utilized YASA and internal tools for consistent annotation workflow.

2025 - 2025

Real-Time Robotic Object Sorting Dataset (Robotic Vision)

ImageObject Detection
Collected and meticulously annotated a custom dataset of fruits (apples, bananas, oranges) using LabelImg to train a YOLOv5s object detection model for an autonomous 4-DOF robotic arm. Ensured tight bounding box annotations across multiple object orientations and used pictures of various lighting setups, which allowed the custom-trained model to achieve a 95% Mean Average Precision (mAP) for real-time pick-and-place sorting.

Collected and meticulously annotated a custom dataset of fruits (apples, bananas, oranges) using LabelImg to train a YOLOv5s object detection model for an autonomous 4-DOF robotic arm. Ensured tight bounding box annotations across multiple object orientations and used pictures of various lighting setups, which allowed the custom-trained model to achieve a 95% Mean Average Precision (mAP) for real-time pick-and-place sorting.

2025 - 2025

Education

O

Obafemi Awolowo University

Bachelor of Science, Electronic and Electrical Engineering

Bachelor of Science
2020 - 2026

Work History

A

African Centre Of Excellence Design Studio

Robotics Engineering Intern

Ile-Ife
2025 - Present
S

Semiconductor Research Group

Student Intern

Ile-Ife
2024 - Present