Patrick Langechuan Liu, Phd Email & Phone Number

Santa Clara, CA, US

Santa Clara, CA, US

Delivering best-in-class ADAS solution with the speed of light.

Bellevue, WA, US

Vision, Language, Robots.

Guangzhou, Guangdong, CN

• Lead of Xpeng's AI Team responsible for camera and lidar perception, camera-lidar fusion, machine learning-based prediction, machine learning infra, and offline perception for automatic annotation. Built and led an.
• Modified transformers module for deployment on Nvidia’s Orin platform to achieve 13x speedup.
• Built 4D Autolabel pipeline for video clips with superhuman accuracy and 2000 clips/day throughput.
• Dynamic XNet to estimate the position/velocity of dynamic agents to reduce dependency on radars/lidars
• Static XNet to provide an accurate estimation of road geometry to reduce dependency on HD mapCreated engineering roadmaps and drove end-to-end delivery of legacy perception stack in Xpilot, Xpeng’s mass-production.
• 360-degree late fusion pipeline with 2.5D perception features. Key enabler for highway solution NGP.

Guangzhou, Guangdong, CN

Camera, Lidar and Radar Perception at fast-growing autonomous driving startup.- Monocular 3D object detection for highway driving- Full perception stack for autonomous parking- 3D object detection, traffic light and road marking recognition for urban driving- Lidar object detection with geometry priors and deep learning- Camera-radar early.

San Diego, CA, US

• Continuing the pursuit of my passion in deep learning, I returned to the field of medical imaging. My daily work involved research and application of state-of-the-art deep learning techniques in image classification.
• Led algorithm development for 3 computer aided diagnosis (CAD) products with deep learning and computer vision techniques
• Delivered algorithm pipeline for a breast cancer screening system with multiple cascaded deep convolutional neural networks (CNNs) to perform density classification, landmark localization, lesion detection and.
• Developed algorithms to identify, localize and quantify the acuity of 10+ thoracic diseases on chest X-ray images
• Initiated algorithm development for rib fractures detection with automatic rib counting and labeling in 3D CT scans
• Improved production pipeline for algorithm deployment by 10x speedup

Oberkochen, DE

• I was one of the first members in a new data analytics team in the Semiconductor Process Control Solutions (PCS) business group.
• Optimized and automated existing 3D measurement workflow based on image analysis software (Volume Graphics), and performed gauge studies to ensure repeatability and reproducibility.
• Developed algorithms for analyzing electron microscope images of memory storage device. Conducted research on machine learning and deep learning algorithms for automatic defect detection and segmentation of 3D.

Shelton, CT, US

• Acting as Senior Detector Physicist in PerkinElmer's Medical Imaging division (now Varex Imaging), I was responsible for the development of X-ray detectors based on amorphous silicon (a-Si) and CMOS technologies.
• Automated X-ray detector defect analysis using deep learning with TensorFlow library, increasing specificity from below 10% to above 80% while maintaining extremely high sensitivity above 99%
• Developed machine learning models based on logistic regression and deployed it in a GUI tool to perform defect analysis in production
• Optimized detector test procedures and cut testing time by 60% through analysis of past test data in Python
• Performed quantitative analysis of x-ray detector performance and translate detector performance requirements to process, component and design requirements: Improved procedures, established criteria and developed.
• Simulated and developed optimized scintillators for various x-ray imaging applications with energies ranging from keV to MeV. Lead engineering effort to increase X-ray detector spatial resolution by 80%+ by adopting a.

Phoenix, AZ, US

• Selected as Fellow from 1800+ applicants (less than 2% acceptance rate) to participate in an intensive two month data science fellowship program.
• Constructed and analyzed a social network among celebrities in New York City using data scraped from more than 1200 webpages.
• Identified the most common type of restaurant violations for different cuisine types based on analysis of more than 500,000 restaurant inspection records in New York City using SQL
• Performed sentiment analysis through construction of bag-of-words and bigram models based on more than 1 million Yelp reviews using Python’s Scikit-learn library
• Predicted business star-rating through linear regression of the features extracted from more than 30,000 Yelp records
• Performed analysis on the entropy level of the English and Thai languages based on all wikipedia pages in Simple English and Thai; Revealed relationships among Wikipedia pages by analyzing the statistics and network of.

Ann Arbor, MI, US

• Research Assistant working on the optimization of megavoltage (MV) X-ray detector for radiotherapy imaging.
• Proposed and optimized the first kV/MV dual energy X-ray detector based on thick, segmented crystalline scintillators
• Pioneered design optimization of megavoltage x-ray detectors for portal imaging and MV cone-beam CT for image guided radiotherapy (IGRT)
• Designed and conducted numerical modeling of detectors through large-scale parallel Monte Carlo simulation on a cluster with more than 800 cores
• Invented a hybrid modeling framework to significantly reduce simulation time from 100 million down to 30 CPU hours (3 million times speedup)
• Designed and implemented 3D tomographic reconstruction and imaging performance metrics analysis using MATLAB

Boston, Massachusetts, US

• Workshop for Advanced Degree Candidates
• Selected as one of the only two participants from University of Michigan
• Collaborated with trainees from various cultural backgrounds on business case analysis and presentations

US

• Selected as one of the only two participants from University of Michigan
• Exposed to and obtained the problem-solving methodology in consulting industry
• Team-worked through practices of mini-cases, group discussion and presentation

北京, Beijing, CN

• Awarded President's Fellowship for Undergraduate Research at Peking University
• Conducted theoretical studies on metabolic networks of eukaryotic and prokaryotic microorganisms
• Developed strong analytical and communication skills through academic research and presentation

Bachelor Of Science (B.S.), Physics

Phd, Physics