Resume for SDE/MLE
Resume for DS/DA
bill.ych.jobs@gmail.com

Messages for Recruiters/Hiring Managers/Professors

  I am Yan-Cheng (Bill) Hsu, a Software Development Engineer with specialized expertise in AI and machine learning, drawing from a robust academic foundation with a Master’s degree from UC San Diego. At Amazon’s Artificial General Intelligence Org, I have been instrumental in engineering secure SageMaker solutions for complex LLM/VLM training, showcasing my capacity to manage extensive datasets with heightened security considerations.

   My tenure at Amazon Alexa Org exemplifies my ability to streamline onboarding processes, achieving a remarkable 90% reduction in integration time through secure backend innovations. This feat reflects my dedication to operational excellence and cybersecurity in the development of AI infrastructures.

   I have made significant strides in AI research, highlighted by my authorship of a publication in the 'Sensors' journal, where I introduced an advanced statistical feature selection algorithm for deep learning applications in vital sign monitoring. Furthermore, my expertise in time series analysis is showcased through my development of self-supervised time series transformers, a work that I have presented at the IEEE APSIPA ASC 2023 conference. This pioneering research contributes to state-of-the-art methodologies in unsupervised learning, propelling advancements in the processing and analysis of temporal data.

   Seeking to leverage my full-stack development skills, cloud computing proficiency, and innovative research background, I am poised to tackle new challenges in AI, aiming to deliver solutions that seamlessly blend technological sophistication with unwavering security and reliability.

• Event 1: Worked as a Software Research Assistent for NCU MLBR Lab
• Event 2: Published on the international journal Sensors as the 1st auther.doi
• Event 3: Worked as a Software Engineer Intern at Wiwynn Inc.
• Event 4: Published my single-page portfolio on AWS Service
• Event 5: Worked as a Software Dev Engineer Intern at Amazon Alexa Org
• Event 6: Earned my Master's degree in Computer Scinece from University of California San Diego
• Event 7: Worked as a Deep Learning Applicaiton Researcher at Swartz Center for Computational Neuroscience
• Event 8: Worked as a Full-Stack Developer at University of California San Diego
• Event 9: Published on the IEEE conference. doi
• Event 10: Worked as a Software Dev Engineer at Amazon Artificial General Intelligence Org - Secure A.I. Foundations

Generalized Deep Neural Network Model for Cuffless Blood Pressure Estimation with Photoplethysmogram Signal Only
Hsu, Yan-Cheng; Li, Yung-Hui; Chang, Ching-Chun; Harfiya, Latifa N. Sensors 20, no. 19: 5668.

• I developed a deep-neural-network model for estimating blood pressure, innovatively integrating a statistical method for feature selection. This enhanced the value of features extracted from Photoplethysmography (PPG) signals.
• The proposed model attained cutting-edge performance in 2020, satisfying both the Association for the Advancement of Medical Instrumentation (AAMI) and the British Hypertension Society (BHS) standards for blood pressure measurement devices.

On the Optimal Self-Supervised Multi-Fault Detector for Temperature Sensor Data
Latifa Nabila Harfiya; Yan-Cheng Hsu; Yung-Hui Li; Jia-Ching Wang Sensors 20, no. 19: 5668.

• Implemented Developed self-supervised time series transformers, securing state-of-the-art performance on diverse temporal datasets.
• Presented findings orally at the IEEE APSIPA ASC 2023 conference, highlighting innovative approaches in time series analysis.

Project I: Olympus Models: Large Language/Visual Models (LLM/LVM) Tooling and Training Optimization on Neurons

• Designed and delivered secure SageMaker solutions for AGI scientists, facilitating the training of LLM/LVMs with 10B+ parameters using over 1TB of data. Optimized processes with pre-compilation, caching, and distributed training capabilities.

• Contributed to the understanding, maintenance, and development of authorization and authentication systems for the ML platform, reinforcing data security for AGI projects and ensuring the safe handling of large datasets.

Comcom Website: Command Line Tools on The Web

• Developed, packaged, and deployed a decoupled web application on AWS EC2, using Linux shell scripts and Docker images. This platform facilitates online code execution and sharing across various languages, eliminating the need for local package installations.
• Built a multi-threaded backend with a SQL database and file-sharing system, capable of managing race conditions, using Flask and Django and created RESTful APIs for request handling.
• Built the frontend with React.js and Node.js, integrating event, state, and proxy management.

Alexa Secure AI Platform: Sensai Self-Service Onboarding Platform

• Created and launched a decoupled web application on AWS Lambda and Cloudfront. The platform facilitates secure onboarding of apps and APIs for scientists on the Sensai Platform.
• Developed modules in backend such as Auto-Verification (which transfer various encrypted identifications via cookies and session data), Canaries, Access Control, and Monitoring using AWS-CDK and AWS-SDK and streamlined the onboarding process for the Alexa Org team, reducing app/API integration time from 4 hours to 15 minutes - a 90% reduction.
• Refined the existing webUI to augment the self-service capabilities of the onboarding system.

Prometheus Infrastructure Testing Data Analysis and Software Toolkit Development

• Established a prototype data pipeline for production line testing data analysis.
• Developed three comprehensive Python packages for efficient data collection, alignment, and analysis from various sources including temporal infrastructure's hardware data (Prometheus) and production line testing databases.
• Streamlined the scope of production line performance enhancement by approximately 66% through effective information filtering and dimensionality reduction.

Deep Neural Network Predictor by Introducing a new Feature Selection Algorithm

• Designed and implemented a deep learning model for blood pressure estimation, incorporating data preprocessing, neural network selection, and a novel physiological feature selection algorithm.
• Enhanced accuracy by ~1.8x and expanded data incorporation by ~6x compared to the predecessor model, achieving a Mean Absolute Error (MAE) of 2.73 mmHg across 2.5M+ cardiac cycles from 9000 patients.
• Published this groundbreaking work in the international journal, Sensors 20.

Template: this