SIO International Wisconsin Inc.⦁ Involved in designing the production process (Design for Automation) to transfer technology from Taiwan to a new manufacturing plant in Wisconsin. Cooperated with Engineering and Manufacturing department to reduce cost and space and identified potential process improvements of the line design while meeting the customer requirements.⦁ Collaborated with the Product Design team to establish the Control Plan and protocols for the production process, utilized FMEA knowledge from ISO/TS 16949 to identify potential failure and preventative action.⦁ Learned the 8-Discipline Problem Solving method and studied cases on the current production line with 5W1H and fishbone diagram to analyze the actual root cause of the product defect.

⦁ Create protocols using the knowledge of USP.1207 and evaluate Container Closure Integrity to support Change Control, product launch, and method qualification of pipeline and commercial products using a qualified vacuum decay method.⦁ Validate the previously developed HVLD testing method and try to develop the new testing method for improvement.⦁ Performed studies on the vacuum decay testing machine and utilized Minitab to explore the results to improve and optimize testing conditions.

Porto Seguro’s Safe Driver Prediction Project⦁ Data Preprocessing with One Hot Encoding and Label Encoding on categorical features, and used XGBoost and LightGBM for feature importance analysis.⦁ Manual tuned one parameter at a time with cross validation followed by using automated tuning with Bayesian Optimization.⦁ Used two levels stacking with best parameters founded to get the normalized gini result of 0.2889NLP Sentimental Analysis Project for Amazon Review⦁ Preprocessed Amazon review raw dataset with tokenization, lemmatization, and tf-itf.⦁ Used Naïve Bayes and Linear SVM to build and find the best model to be used in web application for sentiment analysis.⦁ Linear SVM was found to be the best model with accuracy of 0.8395 and fastest run time.Risk Analysis of Lending Club Loan⦁ Collect dataset from Lending Club website using APIs and perform exploratory data analysis. ⦁ Performed feature engineering with 100+ features.⦁ Trained Extreme Gradient Boosting (XGBoost) model with historical data to identify high-risk loan.⦁ Best AUC scores 0.7 founded on testing dataset from XGBoost with parameter tuned.

⦁ Synthesized Fe2GeS4 thin film transistor that has good electron mobility and suitable energy band gap using ink-jet printing technique. X-ray diffraction was used to identify the crystallinity of sulfurized thin film. Scanning Electron Microscope and Atomic Force Microscope were used to study the film morphology.⦁ X-ray diffraction results show that thin films produced at low temperatures might be amorphous and have crystalline present at temperature 450C or higher. SEM and AFM analyzed film morphology; films are shown rough surface at high sulfurization temperature, and smooth surface with better film quality at a lower temperature.

⦁ Created PVDF thin film containing Nanocrystal cellulose for two-dimensional audio speakers, which shown this can generate sound by applying a current to the film to create stress.⦁ Fabricate films with different ratios of crystalline nanocellulose in solution. Tests were performed to investigate the effect of the ratio of crystalline nanocellulose in solution and the thickness of films. These parameters were analyzed by looking at the volume and sound quality of the speaker, also the frequency range of produced sound.

⦁ Coated Ta2O5 and SiO2 on the glass substrate layer by layer using an ion-assisted deposition technique.⦁ Learned the IAD coating process for WDM telecommunication and what to do when coater malfunctions.