Scoring half a billion consumers for largest banks in three countries.
Our team member was the core part of a Fintech’s big-data and machine learning platform, with advanced credit modelling to predict borrower’s future credit worthiness.
The Fintech startup now works with largest bank in three countries and score nearly half a billion consumers and is on track to score a billion consumers by the end of 2019.
With 11 years of experience in software engineering and 6 years in big-data, data engineering, our team member was in charge of the whole aforementioned start-up’s big-data system from its original creation.
Deep learning model for detecting dangerous cardiac symptoms/diseases which increase the risk of strokes, heart failure and other heart-related complications.
The deep learning model with cutting-edge architecture achieves both accuracy score and f1-score of 93%, which demonstrates a significant improvement in terms of accuracy comparing to the model in production at the time (accuracy & f1-score = 86%), a massive 7% improvement.
Removing the need for cardiologists and doctors to manually review entire length of patient’s ECG records, which sometimes may last for 7 days. Therefore, reducing doctor time, and headcount needed for such task, by approximately 24 times.
Recommendation System for a multitude of mobile application services:
1.Chatbot Framework: to build chatbots upon
2.Virtual Assistant: for internal use of a Bank
3.Virtual Assistant: for onboarding and training Customer Service Reps
4.Open-domain Chatbot
Detecting bad users (reduce over 99% phishing users and prostitute users)
As a mobile app grows, more and more phishing users and prostitute users use the app as a vehicle for their business or tricking other users into certain interactions that might compromise their digital safety, bank account details and confidential personal information.
A machine learning was designed to detect bad users and ban them from using the services, thus, bring better user experience and quality of services to the user base.
1. Mining users relationship (detected Family Members and Colleagues of over 20 million users).
2. Design and develop the algorithm for friend suggestion.
3. Analyze users behaviors and interests.
Plan and propose different approaches, data engineering, data analytics and machine learning solutions
Understanding and documenting business problems. Research existing data infrastructure, platform and conducting feasibility study. Architecting and designing the solutions.
Developing customized, bespoke machine learning, deep learning models that answers the problems our clients have.
Deploy and automate machine learning models. Ensure the deployed software/model is robust and scalable, and requires mininum human intervention.