Matthew Angus

Bringing research into production

Work Experience

Feb ‘24 - Present Principal Computer Vision Engineer, Humanising Autonomy, London UK

  • Regularly guided the business with technology decisions, such as client-facing feature deprecation strategy
  • Productionised multi-task learning training, testings, model conversion, and model release
  • Created forward collision warning evaluation, and metrics tracking
  • Transitioned from GOCD to GitHub actions and Championed adoption of Conan
  • “Innovation day” projects widely used. For example, automated SDK python bindings and camera calibration web-app

Oct ‘22 - Feb ‘24 Senior Computer Vision Engineer, Humanising Autonomy, London UK

  • Architected MLOps workflow from the ground up, automating reporting and
  • Built a multi-task learning framework, and trained Object detection, lane detection, and camera obstruction tasks jointly
  • Contributed to production C++ codebase, such as implementing an omnidirectional camera model
  • Led team of one mid and two junior level engineers to overhaul the company’s MLOps
  • Automated the creation of a synthetic collision dataset with 300 videos from 20 scenarios dataset using CARLA simulator

Jan ‘20 - Feb ‘22 Computer Vision Engineer, Humanising Autonomy, London UK

  • Training YOLO object detection models with Darknet, writing custom augmentation and LR schedules in C++
  • Leveraged open source scene reconstruction libraries combined with semantic segmentation to generate and label maps

May ‘18 - Aug ‘18 Research Engineer, NXP Semiconductors, Ottawa Canada

  • Extended state of the art world models for autonomous driving.

  • Implemented and trained deep learning models, such as variational auto-encoders and RNNs, in TensorFlow using the Carla driving simulator.

  • Researched independently, providing key milestones to co-workers.

Jan ’15 - Aug ’16 Application Developer, Canadian Natural Resources Limited, Calgary Canada

  • Consulted on architecting the data and business layers of a responsibility management system that tracked responsibilities globally, that extracted a hierarchy from a relational database.

  • Helped translate internal customer requirements into action items for our development team.

  • Championed and implemented bug tracking best practices across our development team, previously there was none.

May ’12 - Dec ’13 Security Analyst, Canadian Natural Resources Limited, Calgary Canada

  • Took initiative and automated various manual processes through the use of scripts, saving the team about 3 hours per week.

  • Performed within the company’s workflow to deliver access to various applications.

Education

Sept ‘16 - July ‘19 Master of Mathematics in Computer Science, The University of Waterloo. – 3.98/4 Major GPA

  • Semantic segmentation and pixel-level out-of-distribution detection.

  • Wrote custom TensorFlow unpool operation with CUDA GPU implementation, with \(2\times\) speed up at inference time.

  • Created largest public synthetic segmentation dataset to date using GTAV (1 million+ images).

  • Multiple courses and course projects in machine learning.

Sept ’11 - April ’16 Bachelor of Science in Computer Science, The University of Calgary, Minor in Pure Mathematics, Concentration in Algorithms and Complexity Theory. – 3.68/4 Major GPA

  • Sought out Dr. Rob Kremer to collaboratively research ontology inference for a semantic knowledgebase where I decreased the runtime of our inference algorithm by 97% to achieve a runtime in \(O(x^n)\).

  • Two semester exchange at the University of Western Australia where I was able to grow personally, academically and culturally by studying and traveling in Australia and surrounding countries.

Awards and Competitions

2016 Math Domestic Graduate Student Award

2016 Graduate Research Studentship Award

2016 University of Calgary Faculty of Science Dean’s List

2016 Intelligent Ground Vehicle Competition Rookie of the Year

2016 Intelligent Ground Vehicle Competition 8th Place, out of 24

2016 IEEE Sumobots Competition

2015 Alberta Collegiate Programming Contest

2015 IEEE Minibots Competition

2013 International Studentship Competition Award

2012 Alberta Collegiate Programming Contest

Technical skills

Basic LATEX, Haskell

Intermediate C++, Python, TensorFlow, Test Driven Development, Git, OpenCV, Linux

Advanced Java, Matlab, SQL, C#

UCalgary Autonomous Robotics Club

2013 - 2016 Lead Software Developer

  • Architected and implemented over 50% of the core system includingthe real-time image processor for vehicle navigation.

  • Integrated fuzzy controller with systems data streams (GPS, Lidar, IMU)

2012 - 2013 Software Developer

  • Experimented with motion planning algorithms such as A* search basedmotion planner and the distance transform to find open space.

2013 - 2016 Club Executive

  • Co-managed ARC where I helped managed our budget and recruitment on campus.

  • Worked on a community outreach program to present our robot to middle school students to help gain interest in robotics and computer science.

  • Organized and managed ARC’s first entry into the Intelligent Ground Vehicle Competition.

Interests

  • Tough Mudder

  • Raspberry Pi

  • Rock Climbing



Download PDF Version