Personal details
Osasu U.
Based in: 🇳🇬 Nigeria
Timezone: West Central Africa (UTC+1)
Summary
7+ years expertise in advanced machine learning and deep learning, I specialize in optimizing data processing and driving impactful decision-making. I’ve led projects resulting in significant reductions in model deployment and data loading times, alongside substantial capital savings. From spearheading innovative solutions to orchestrating efficient ETL pipelines, I consistently deliver high-impact results.
Work Experience
Senior Data Scientist
Okra Inc. | Mar 2020 - Present
Python
Machine Learning
ETL
TensorFlow
Data Engineering
OpenAI
AWS (Amazon Web Services)
Scikit-learn
- Spearheaded the development of a question and answer system utilizing LLMs and RAG for API documentation queries, resulting in a 15% reduction in API support requests.
- Orchestrated ETL and ELT pipelines in AWS, extracting data from SQL, No-SQL, and third-party sources into Redshift, reducing loading time by 40% for real-time analytics.
- Enabled ML models adapt to evolving data by building a continual learning system incorporating performance monitoring, data curation and QA, drift detection, and retraining strategy.
- Authored company data science standard documents, including onboarding guides, data version control protocols and deployment strategies ensuring consistency and operational efficiency.
- Compiled GPU-based deep learning models to run efficiently on CPU using AWS Neuron SDK and Inferentia accelerators, reducing inference time by 40% and achieving 20% cost savings.
- Leveraged LSTM and BERT models for text classification and Named entity recognition to generate detailed analysis on financial data, boosting data-driven decisions by 20% in 6 months.
- Performed unsupervised credit risk analysis utilizing mathematical and statistical models, leading to a 15% reduction in default rates and improved lending portfolio performance.
Data Scientist
BoroMe Inc | Feb 2018 - May 2020
Python
Computer Vision
JavaScript
Docker Compose
TensorFlow
Scikit-learn
- Led the creation of scalable ML APIs and infrastructure utilizing Flask and Docker, resulting in a 30% increase in loan application submissions over 5 months, enhancing user experience.
- Innovated Computer Vision models for precise image quality detection, enhancing facial recognition accuracy and facilitating ID card extraction using OCR with 99% accuracy.
- Lowered loan default rate and improved capital savings by 25% leveraging advanced statistical analysis and machine learning models for credit risk analysis.
Education
Nile University
Master's degree・Computer Science
Jan 2019 - Oct 2021
Covenant University
Bachelor's degree・Management Information System
Sep 2011 - Jul 2015
Personal Projects
2018
Python
TensorFlow
In this project, a content based image retrieval problem is approached using unsupervised machine learning techniques i.e Auto-Encoders.