Emmanuella Sule

Emmanuella Ohunene Sule ML Engineer (- | Abuja, Nigeria Email | Github | LinkdeIn | Blog SUMMARY A detail-oriented Machine Learning and AI Engineer. Proficient in vision modeling, time series forecasting, and deep and machine learning models to solve real-world problems. My approach is highly agile, transparent, and structured. I tackle project development from zero to hero and have a strong commitment to quality. EDUCATION B.Eng, Electrical and Electronic Engineering, Federal University of Technology, Minna.​- ​ Relevant Courses: Probability, Statistics, Linear Algebra,​ Calculus, Analog and Digital Electronics.​ ​ TECHNICAL SKILLS AND SOFTWARE PROFICIENCIES Languages: Python, R, SQL, MySQL Developer Tools: Git, GitHub, GitLab, CLI tools, VS Code, Jupyter Notebooks, Google Colab Libraries: PyTorch, Keras, scikit-learn, LightGBM, XGBoost, NLTK, pandas, NumPy, HuggingFace Transformers, OpenCV, YOLO, Ultralytics Databases: MySQL, BigQuery, SQLServer WORK EXPERIENCE Acid Integrations - Canada​ ​ ​ ​ ​ ​ ​ April 2025 to date Junior ML Engineer ●​ Collaborated with cross-functional teams to engineer and deploy machine learning solutions across various initiatives. ●​ Developed an OCR pipeline for accurate jersey number recognition, contributing to a professional sports video analysis project. Acid Integrations - Canada​ ​ ​ ​ ​ ​ ​ November 2024 - April 2025 ML Intern ●​ Completed a 6-month AI Development Internship, gaining hands-on experience in AI/ML development. ●​ Collaborated with Senior Engineers to build and refine machine learning solutions for real-world scenarios. ●​ Strengthened practical knowledge of the end-to-end ML pipeline and core frameworks, with a focus on PyTorch. ​ ​ Verse Telecom LTD - Nigeria​ ​ ​ ​ ​ ​ ​ September 2023 - January 2024 Network Engineering Intern ●​ Collaborated with network engineers and installers on deploying fiber optic internet access to customers and businesses. ●​ Configured an Optical Network Terminal (ONT) and prepared a simplified technical report on the process. ●​ Managed customer data collection and entry, ensuring accuracy and integrity. PROJECTS Time-Series Analysis on IoT Sensor Data | Python, Pytorch ●​ Engineered a seq2seq model with Bahdanau attention to forecast IoT sensor data, improving prediction accuracy by 18% compared to baseline models. ●​ Enhanced reliability in sensor fault detection by applying Borderline-SMOTE for balanced classification. ●​ Leveraged explainable AI to provide insights into sensor fault predictions Emotion Recognition in Infants | Python, Pytorch ●​ Developed a ResNet-18 model from scratch to classify infant emotions, achieving 92.29% accuracy. This scalable solution could be used for emotion monitoring, supporting healthcare professionals in early developmental assessments. UNet Image Segmentation | Python, Pytorch ●​ Implemented a custom UNet architecture for binary semantic segmentation. ●​ Applied Soft IoU (Jaccard) Loss, which outperformed Dice Loss, aligning optimization more closely with the evaluation metric. Face Recognition System to Record Class Attendance | Python, OpenCV ●​ Designed a low-cost, real-time classroom attendance system using ESP32 camera and LBPH algorithm. ●​ Achieved 92% face recognition accuracy and cut manual attendance tracking time by over 70%. ●​ Delivered a companion mobile app for remote access to the attendance database, increasing administrative efficiency. Customer Segmentation and Churn Prediction | Python, Scikit-learn ●​ Developed a machine learning model that identified at-risk customers with over 93% accuracy. ●​ Segmented customers into distinct categories using K-means clustering, providing businesses with data-driven insights to tailor marketing and customer success efforts. Analysis of Divvy Bikes | R, Data analytics ●​ Conducted an in-depth analysis of Divvy’s historical data, identifying key usage patterns that informed business strategy and helped to optimize marketing efforts for attracting new annual members. ●​ Developed an interactive dashboard to present key insights to stakeholders, enabling 25% faster decision-making on product and marketing initiatives.