Machine Learning and Data Science Intern at IVANI (Smart home device company)   

                                                                                                    May 2016 - Aug 2016                                                               

• Develop presence sensing algorithm implemented with machine learning models in Python

• Surpass the main algorithm so the company implements my work and puts it under 2 patents and NDA

• Automate sensor data collection, aggregation; data analysis and visualization

• Work directly with senior engineers and demo to the board and clients

I lead the data intern team of 2 and interacted with sensors to collect signal data for designing a human presence sensing algorithm. I statistically analyze entropy from data, normalize and use different filters to clean the noise and experiment to design best feature combination for classifiers. After experimenting different models like CNNs, creating different evaluation metrics and , due to lightweight requirement, I chose to use SVC ( by liblinear and libsvm) and further fine-tune its hyperparameters like kernel functions. My final algorithm successfully surpassed the company’s main algorithm and getting funded by IVANI for 2 ML related patents. I had the chance to demo in front of the board, trustees and potential major clients. It was my first time to bring ML into real application world for end users.