Principal Engineer

• Sep 2023 - Present

Leading a team of engineers and scientists in the development of next-generation lateral flow devices that use computer vision to detect diseases and other health conditions more accurately and efficiently.

Software Engineering Manager

• Mar 2021 - Present

• Designed and implement software architecture for the new real-time PCR Instrument, and led the global software team to commercialize the product. • Led the lateral flow reader product, developed a new image processing technique, and achieved 400% more throughput and 800% fast than the competitors. •Identify the positive patents in covid-19 tests through machine learning. • Conduct employee reviews and performance plans for Engineers. • Prepare project execution plans, schedules… Show more • Designed and implement software architecture for the new real-time PCR Instrument, and led the global software team to commercialize the product. • Led the lateral flow reader product, developed a new image processing technique, and achieved 400% more throughput and 800% fast than the competitors. •Identify the positive patents in covid-19 tests through machine learning. • Conduct employee reviews and performance plans for Engineers. • Prepare project execution plans, schedules, and budgets while participating in all project work phases from conceptual design to post-construction field checkout. • Direct engineering activities and coordinate resources to achieve project goals. • Implemented the agile development process and educated business members and stakeholders on the agile process's benefits. •CE/EUA/FDA application preparation.

Senior Systems Software Engineer

• Mar 2020 - Mar 2021

Develop the software and algorithm system for molecular diagnostics PCR device

Computer Vision & Machine Learning Engineer

• Nov 2019 - Mar 2020

Ordinary work includes data mining anomaly detection, signals & patterns analysis, data modeling, decision trees/forests, clustering analysis. Develop new models and initiatives that help the business make the best decisions to maximize the usage of the system. Join the team for developing AI algorithm for the new generation 100x oil optical auto scanning microscope. Such as training the algorithm to detect the different sample shapes, putting some focusing points on the sample automatically… Show more Ordinary work includes data mining anomaly detection, signals & patterns analysis, data modeling, decision trees/forests, clustering analysis. Develop new models and initiatives that help the business make the best decisions to maximize the usage of the system. Join the team for developing AI algorithm for the new generation 100x oil optical auto scanning microscope. Such as training the algorithm to detect the different sample shapes, putting some focusing points on the sample automatically. The San Diego engineer team was closed on March 2020 due to Covid-19 pandemic Show less Ordinary work includes data mining anomaly detection, signals & patterns analysis, data modeling, decision trees/forests, clustering analysis. Develop new models and initiatives that help the business make the best decisions to maximize the usage of the system. Join the team for developing AI algorithm for the new generation 100x oil optical auto scanning microscope. Such as training the algorithm to detect the different sample shapes, putting some focusing points on the sample automatically… Show more Ordinary work includes data mining anomaly detection, signals & patterns analysis, data modeling, decision trees/forests, clustering analysis. Develop new models and initiatives that help the business make the best decisions to maximize the usage of the system. Join the team for developing AI algorithm for the new generation 100x oil optical auto scanning microscope. Such as training the algorithm to detect the different sample shapes, putting some focusing points on the sample automatically. The San Diego engineer team was closed on March 2020 due to Covid-19 pandemic Show less

Staff Systems Engineer

• Sep 2015 - Nov 2019

Generated V&V protocols, design of control, and design risk assessment analysis for the ultrasound diagnostic instrument. Minimize the risk of biological side effects associated with the use of the medical ultrasonic diagnostic machine by devising and implementing new predictive models and initiatives, enabling the user to make the best decisions to maximize the received signal with the least power output. Utilize advanced analytical techniques and statistical programming languages to identify… Show more Generated V&V protocols, design of control, and design risk assessment analysis for the ultrasound diagnostic instrument. Minimize the risk of biological side effects associated with the use of the medical ultrasonic diagnostic machine by devising and implementing new predictive models and initiatives, enabling the user to make the best decisions to maximize the received signal with the least power output. Utilize advanced analytical techniques and statistical programming languages to identify patterns in user operational behaviors, optimize the operational conditions for the user. Design and deliver an automated algorithm that accurately forecasts the harmfulness of millions of operational scenarios to streamline the decision-making and execution process of the R&D related to diagnostic medical equipment. Allow the company to know ultrasound mechanic and thermal indices by aiding the use of above mentioned FDA approved model on products. Key Achievement:  Provide quality assurance of 3M sets of measurement data, along with overseeing anomaly detection, dimension reduction, and classification analysis, and build a real-time predictive model of ultrasound safety. FDA approved our predictive model.  Developed patient database and integrated it into the system of diagnostic ultrasound products, along with user experience of SQL script for IBM database on cloud.  Participates in the design of instruments to ensure they meet the requirement of related regulations, such as ISO 13485, 14971, and IEC 60601, help the company to pass FDA 510k application.  Created efficient algorithms for acoustic signal processing and configuration of the ultrasonic diagnostic system.

Senior System Engineer

• Sep 2012 - Sep 2015

System Engineer

• May 2011 - Sep 2012

RESEARCH ASSISTANT

• Aug 2004 - Dec 2009

Data mining and predictive modelling. • Data mining. Processed 200 million raw galaxies and sense candidates by coordination, r-band magnitude, and red-shift; efforts involved association rule learning, anomaly detection, cluster analysis and dimension reduction (split 3-D space into 2-D slices). • Data classification. Akin to the Support Vector Machine technique; partnered with team to develop Metric Space Technique adapted to perform a space-scale morphological analysis… Show more Data mining and predictive modelling. • Data mining. Processed 200 million raw galaxies and sense candidates by coordination, r-band magnitude, and red-shift; efforts involved association rule learning, anomaly detection, cluster analysis and dimension reduction (split 3-D space into 2-D slices). • Data classification. Akin to the Support Vector Machine technique; partnered with team to develop Metric Space Technique adapted to perform a space-scale morphological analysis by altering galaxy point distributions, a discrete variable, with smoothing Gaussian function, thus giving quantitative structural information (voids or clusters) on all size scales. • Predictive modelling & cross validation. Based on the results of data classification, we suggest a improved universe evolution model, and we also utilized N-body simulation data for cross validation with predictive modeling; test whether the predicted new phenomena in future evolution of universe consistent with the simulation. Developed data mining tool and data mining. • Develop data mining tool. Conceived Hexagon Cell Technique, comparing with traditional square 2-D method to divide 2-D point distribution into units, hexagon provides more sides 6 sides, than square 4 sides; could be sensitive to environment and provide extra information about space geometry, topology, data clustering, and nearest neighbors; this new data mining tool helped us to quanitfy the complicated 2-D structure of the universe. • Data mining. After data mining for 0.6 million raw galaxies in Two Micron All-Sky Survey, discovered low-density regions have significantly less fiber structure and that they are similarly less clustered than high-density regions; differences could be caused by evolution suggesting it might include more things than gravitational instability or caused by original asymmetry of quantum fluctuation in early universe. Show less Data mining and predictive modelling. • Data mining. Processed 200 million raw galaxies and sense candidates by coordination, r-band magnitude, and red-shift; efforts involved association rule learning, anomaly detection, cluster analysis and dimension reduction (split 3-D space into 2-D slices). • Data classification. Akin to the Support Vector Machine technique; partnered with team to develop Metric Space Technique adapted to perform a space-scale morphological analysis… Show more Data mining and predictive modelling. • Data mining. Processed 200 million raw galaxies and sense candidates by coordination, r-band magnitude, and red-shift; efforts involved association rule learning, anomaly detection, cluster analysis and dimension reduction (split 3-D space into 2-D slices). • Data classification. Akin to the Support Vector Machine technique; partnered with team to develop Metric Space Technique adapted to perform a space-scale morphological analysis by altering galaxy point distributions, a discrete variable, with smoothing Gaussian function, thus giving quantitative structural information (voids or clusters) on all size scales. • Predictive modelling & cross validation. Based on the results of data classification, we suggest a improved universe evolution model, and we also utilized N-body simulation data for cross validation with predictive modeling; test whether the predicted new phenomena in future evolution of universe consistent with the simulation. Developed data mining tool and data mining. • Develop data mining tool. Conceived Hexagon Cell Technique, comparing with traditional square 2-D method to divide 2-D point distribution into units, hexagon provides more sides 6 sides, than square 4 sides; could be sensitive to environment and provide extra information about space geometry, topology, data clustering, and nearest neighbors; this new data mining tool helped us to quanitfy the complicated 2-D structure of the universe. • Data mining. After data mining for 0.6 million raw galaxies in Two Micron All-Sky Survey, discovered low-density regions have significantly less fiber structure and that they are similarly less clustered than high-density regions; differences could be caused by evolution suggesting it might include more things than gravitational instability or caused by original asymmetry of quantum fluctuation in early universe. Show less

RESEARCH ASSISTANT

• Sep 2001 - Jun 2004

Developed data mining tool and data mining. • Developed data mining tool. Created Rhombic Cell Technique which compares with traditional cube 3-D methods for dividing 3-D point distribution into units, rhombic dodecahedron provides more faces, 12 faces, than cube - 6 faces; could be sensitive to environment and provide extra information about 3-D space geometry, topology, data clustering, and nearest neighbors; this new data mining tool supported quantitative complicated 3-D structure of… Show more Developed data mining tool and data mining. • Developed data mining tool. Created Rhombic Cell Technique which compares with traditional cube 3-D methods for dividing 3-D point distribution into units, rhombic dodecahedron provides more faces, 12 faces, than cube - 6 faces; could be sensitive to environment and provide extra information about 3-D space geometry, topology, data clustering, and nearest neighbors; this new data mining tool supported quantitative complicated 3-D structure of the universe. • Data mining. Process for 18,000 raw galaxies in IRAS Point Source Catalog Redshift Survey, compared with mock catalogs produced by Monte Carlo method; determined universe is more like loose collections of lakes on land than groups of islands and archipelagos in an ocean. Show less Developed data mining tool and data mining. • Developed data mining tool. Created Rhombic Cell Technique which compares with traditional cube 3-D methods for dividing 3-D point distribution into units, rhombic dodecahedron provides more faces, 12 faces, than cube - 6 faces; could be sensitive to environment and provide extra information about 3-D space geometry, topology, data clustering, and nearest neighbors; this new data mining tool supported quantitative complicated 3-D structure of… Show more Developed data mining tool and data mining. • Developed data mining tool. Created Rhombic Cell Technique which compares with traditional cube 3-D methods for dividing 3-D point distribution into units, rhombic dodecahedron provides more faces, 12 faces, than cube - 6 faces; could be sensitive to environment and provide extra information about 3-D space geometry, topology, data clustering, and nearest neighbors; this new data mining tool supported quantitative complicated 3-D structure of the universe. • Data mining. Process for 18,000 raw galaxies in IRAS Point Source Catalog Redshift Survey, compared with mock catalogs produced by Monte Carlo method; determined universe is more like loose collections of lakes on land than groups of islands and archipelagos in an ocean. Show less