Jack Kwok, Developer in San Francisco, CA, United States
Jack is available for hire
Hire Jack

Jack Kwok

Verified Expert  in Engineering

Deep Learning Developer

San Francisco, CA, United States
Toptal Member Since
May 23, 2019

With 20 years of experience, Jack excels in developing complex software systems and custom deep-learning models for computer vision and natural language processing. He specializes in solving complex problems like 2D and 3D object detection, multi-object tracking, instance segmentation, and document understanding and classification. Jack has trained cutting-edge computer vision models for healthcare and autonomous driving problems, and he now focuses on LLM and generative AI.


Computer Vision, Generative Pre-trained Transformers (GPT), GPT...
Amazon Web Services (AWS), Object Detection, Deep Learning, OpenCV, Python...
Insight Data Science
Keras, Deep Learning




Preferred Environment

Amazon SageMaker, Google Cloud ML, Amazon Web Services (AWS), Scikit-learn, Windows, Linux, PyTorch, Python, MacOS

The most amazing...

...project I have worked on involved building a neural network from scratch to recognize handwritten digits. It led me to pursue a career in deep learning.

Work Experience

Senior Machine Learning Scientist | Tech Lead

2020 - 2023
  • Developed state-of-the-art PyTorch computer vision models for automated clinical diagnosis and anatomical identification from dental radiographs.
  • Built natural language processing (NLP) and visual document understanding (VDU) models for insurance document classification and clinical narrative comprehension.
  • Created a scalable image processing inference pipeline on the Google Cloud Platform using Cloud Functions, Cloud Run, and Google Kubernetes Engine (GKE).
Technologies: Computer Vision, GPT, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Google Cloud Platform (GCP), Image Processing, PyTorch

Staff Software Engineer | Computer Vision and Machine Learning

2017 - 2020
  • Applied deep learning and computer vision algorithms to solve autonomous driving problems.
  • Led the design and development of custom deep learning-based multi-object trackers used by multiple teams to track cars, cyclists, pedestrians, and traffic control elements.
  • Designed the initial production release of the active learning data pipeline for continuous machine learning model improvement.
Technologies: Amazon Web Services (AWS), Object Detection, Deep Learning, OpenCV, Python, TensorFlow, PyTorch

Artificial Intelligence Fellow

2017 - 2017
Insight Data Science
  • Developed a Computer Vision system based on a convolutional neural network that automatically annotates post-hurricane flooded roads on satellite imagery.
  • Developed an open-source project on GitHub.
  • Researched and published a blog post about the use of deep learning for disaster recovery.
Technologies: Keras, Deep Learning

Staff Software Engineer

2015 - 2017
  • Helped build LinkedIn's video platform from scratch.
  • Wrote a technical blog post on building a Native video player library for Android.
Technologies: Devices, Mobile, Video Codecs, Java

Software Architect

2011 - 2015
  • Scaled our mobile platform to over a dozen apps across iOS, and Android. This grew the audience over ten times. Trulia mobile apps are top-rated by users, and frequently selected for prestigious editorial features by Apple, Google Android (Editors' Choice, and Staff Pick), Amazon, and Samsung.
  • Developed and built projects for iOS and Android involving geofencing and geospatial data, user data synchronization, cross-platform communication, and natural language processing.
Technologies: REST APIs, iOS, Android

Technical Leader

2006 - 2011
  • Led the architectural designs of Cisco Jabber (an enterprise VoIP app) on Android. Worked with product managers to translate business needs into technical specifications. Worked with cross-functional teams to define requirements, interfaces, and implementation approaches, and led a group of six talented Android developers to build out layered components of the application from scratch.
  • Extended the Blackberry client with innovative voice features, coordinated development activities, and mentored junior engineers.
Technologies: Android

Senior Software Engineer

2004 - 2006
Weathernews Americas
  • Developed the prototype, architecture, design, and implementation of Weathernews LiveLocal. LiveLocal is the US’s first streaming video application bringing videos from local broadcast TV stations to cell phones.
Technologies: Videos, Mobile

Deep Learning for Disaster Recovery

I attempted to automatically annotate flooded, washed out, or otherwise severely damaged roads using state-of-the-art deep learning methods. My goal was to create a tool to help detect and visualize anomalous roads in a simple user interface.

For details, see the blog post:


PyTorch, OpenCV, Scikit-learn, NumPy, Keras, Pandas, SciPy, REST APIs, TensorFlow


Scikit-image, Amazon SageMaker


Computer Vision Algorithms, Deep Learning, Computer Vision, Object Detection, Object Tracking, Image Processing, Machine Learning, Mobile Apps, Linear Algebra, Convolutional Neural Networks (CNN), Devices, Videos, Google Cloud ML, Video Codecs, Recurrent Neural Networks (RNNs), Natural Language Processing (NLP), GPT, Generative Pre-trained Transformers (GPT), Computer Science, Artificial Intelligence (AI), 3D Math, Geometry


Python, Java


Google Cloud Platform (GCP), Linux, Windows, Mobile, Android, iOS, Amazon Web Services (AWS), Kubernetes, MacOS

2001 - 2002

Master of Engineering Degree in Computer Science

Massachusetts Institute of Technology - Cambridge, MA, USA

1997 - 2001

Bachelor of Science Degree in Computer Science

Massachusetts Institute of Technology - Cambridge, MA, USA

1997 - 2001

Bachelor of Science Degree in Mathematics

Massachusetts Institute of Technology - Cambridge, MA, USA


Google Cloud Platform Professional Machine Learning Engineer


Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.


Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring