大象传媒

MENU
Faculty of Applied Sciences
School of Computing Science
Search

INDUSTRY TALK SERIES

Industry Talks Series

Matthew Leonard

Date: Monday, October 27, 2025

Time: 5:00 - 6:30 PM PST

Location: TASC 1 9204, Burnaby campus

Talk Title: Software-Defined Storage with Ceph: From Data Centers to Open Source Innovation

Abstract: In this session, we鈥檒l explore the journey from the physical realities of modern data centers鈥攑ower, cooling, and rack constraints鈥攖o the software-defined storage systems that make global-scale computing possible. Using Ceph as our guide, we鈥檒l look at how open source storage provides block, file, and object services at scale, the challenges it faces both technically and in the market, and why it matters for the future of cloud and data-driven applications. Students will gain insight into how industry leaders like Bloomberg deploy Ceph, the trade-offs between traditional and software-defined approaches, and how they can engage with the Ceph community through labs, open source contributions, and career opportunities.

Speaker Info: Matthew Leonard is Head of Compute & Storage Engineering at Bloomberg, where he leads a global team of nearly 100 engineers across three continents. His organization provides Compute-as-a-Service, Storage-as-a-Service, and automation platforms to more than 9,000 Bloomberg developers, powering mission-critical analytics and private cloud systems. His teams manage everything from bare-metal provisioning to large-scale private clouds built on OpenStack and VMware ESX, as well as storage platforms that combine Ceph, vendor appliances, and other software-defined solutions. Matthew also serves as Chairman of the Board of the Ceph Foundation, working with the Linux Foundation and global community to guide the growth and sustainability of one of the world鈥檚 most important open-source storage projects. With a background spanning aerospace engineering, military aviation, research, telecom, and large-scale financial computing, Matthew brings a unique perspective on how diverse technical experiences shape resilient infrastructure and strong engineering organizations. Outside of work, he is an interior firefighter with the East Marion Fire Department in New York, reflecting his commitment to service both in technology and in the community.

Previous Talks

Carlos Alberto da Costa Filho & Mike Wong

Date: Wednesday, September 24, 2025

Time: 4:30 - 6:00 PM PST

Location: TASC 1 9204, Burnaby campus

Talk Title: Solving the DarkVision challenge: finding all safety-critical needles in massive haystacks

Abstract: DarkVision Technologies provides sub-millimetre acoustic imaging to support industrial asset integrity, but analyzing these massive datasets poses significant challenges for automated defect detection. In this talk, we highlight some of these challenges and the machine learning tools we use to address them, focusing in particular on DVT-AD, a ViT self-distillation approach trained solely on normal data. DVT-AD achieves 99.3鈥�% AUROC on MVTec-AD and 99.1鈥�% on our large-scale ultrasound dataset, outperforming all other state-of-the-art methods tested. We will explore in depth why our method surpasses existing approaches and discuss the implications for deploying deep learning in real-world industrial inspection.

Speaker Info: Carlos Alberto da Costa Filho is Research Manager at DarkVision Technologies, North Vancouver, where he leads work on automated ultrasound image processing, anomaly detection, and high-performance deep learning systems. He previously held research positions at Viridien (formerly CGG) in Rio de Janeiro, focusing on large-scale inverse problems and deep learning for subsurface imaging, and at the University of Edinburgh as a Postdoctoral Research Assistant. He earned his Ph.D. in Geophysics from the University of Edinburgh in 2017, following an M.S. in Applied Mathematics from the University of Campinas (2013) and a B.S. in Mathematics from Pontif铆cia Universidade Cat贸lica do Rio de Janeiro (2010).  

Mike Wong is a Machine Learning Scientist at DarkVision Technologies, North Vancouver, where he develops advanced algorithms for ultrasonic and volumetric image analysis, with a focus on unsupervised anomaly detection, self-supervised representation learning, and scalable deep learning systems. His work also includes cloud-based training and inference pipelines and the design of domain-specific neural network architectures for industrial inspection. He previously held research and applied machine learning positions in the financial, retail, and sports analytics sectors, specializing in time-series modeling and computer vision. He received his B.Eng. degree in Computer Science from the Hong Kong University of Science and Technology in 2018.

Vincent Chu

Date: Wednesday, April 30th, 2025

Time: 3:00 - 4:30 PM PST

Location: TASC 1 9204, Burnaby campus

Talk Title: Building Effective Data Teams: Streaming, ETL, and LLM-Powered Reporting 

Abstract: How do small teams succeed in big organizations? This talk explores key strategies for building an effective data team, from quickly ingesting data with streaming architectures to adapting traditional ETL pipelines. We鈥檒l also discuss how large language models (LLMs) can make reporting more accessible while avoiding misinformation. Beyond the tech, we鈥檒l cover how teams can stay agile, collaborate across departments, and design clear interfaces for long-term success. Whether you're pursuing a career in a startup or a large enterprise, in AI or data, this session will provide valuable insights into real-world data challenges and solutions.

Speaker Info: Vincent Chu, VP of Engineering at Visier, is a seasoned engineering leader with extensive experience in building high-performance teams, driving technical innovation, and scaling software systems. Vincent鈥檚 career has spanned both startups and large-scale enterprises, including organizations like Electronic Arts and Apple. He has successfully led teams at companies such as Visier, Later.com, and AgencyAnalytics, where he played a pivotal role in designing and delivering mission-critical platforms.

Shadid Chowdhury

Date: Wednesday, March 26th, 2025

Time: 3:00 - 4:30 PM PST

Location: TASC 1 9204, Burnaby campus

Talk Title: DNS: The Silent Weapon of Cybercriminals 鈥� Leveraging Data and AI/ML to Build DNS-Based Cybersecurity Solutions

Abstract: Cybercriminals are increasingly exploiting DNS to bypass traditional security measures, execute stealthy attacks, and exfiltrate data undetected. In this session, we'll expose the hidden risks within DNS, discuss the challenges of building effective security insights, and demonstrate how AI/ML-driven analytics can detect and disrupt threats before they cause harm. Join us to explore how innovative data strategies and advanced machine learning can transform DNS from a security blind spot into a powerful defense tool. 

Speaker Info: Shadid Chowdhury is a Senior Engineering Manager at Infoblox, where he leads the development of DNS-based cybersecurity products. He holds a Master鈥檚 degree in Distributed Systems from the Royal Institute of Technology (KTH), Sweden. Before joining Infoblox, Shadid worked at Amazon Web Services (AWS), Teradata, SmartBear, and Klarna, gaining deep expertise in cloud computing, big data, and software engineering. He brings valuable insights into building robust, scalable cybersecurity solutions. 

Gordon Niguma

Date: Wednesday, February 26, 2025

Time: 3:00 - 4:00 PM PST

Location: TASC 1 9204, Burnaby campus

Talk Title: Who is Lamar Duckworth, why do your parents need tech support, and should you lend me money? 

Abstract: This tackles three basic questions with real-world examples, that are affected by real-world technology problems. The three problems we tackle today are:

  • 鈥淗i Lamar Duckworth, are you Lamar Duckworth? The reality about Data鈥� 鈥� 鈥淲ho is Lamar Duckworth鈥� is a surprisingly complex question with no easy answers. One reason is because of data quality and understanding how data quality affects our abilities to solve real-world problems. It can lead to losing money, whether its because of fraud, missed opportunities to market to someone or making bad decisions on new products.  
  • 鈥淲hy do your parents ask you for tech support and can鈥檛 beat you at video games?鈥� 鈥� Your parents were built for the 70s and 80s. They can鈥檛 text quickly, but they can type. They can鈥檛 play video games if there鈥檚 more than 4 buttons on the controllers. But the same is true for old technology and why it hinders our ability to progress forward. 
  • 鈥淪hould I lend this person money?鈥� 鈥� If a stranger asked you for money, how would you determine if you should let them borrow it? This will be a group effort to discuss the basic process of lending, how it works in practice, all of the technologies that are required, and the considerations that have to be made.  

Speaker Info: Gordon Niguma is a Director in the Enterprise Architecture team at Coast Capital Savings. He earned his M.Sc. in Computer Science from 大象传媒. Previously, Gordon was a Manager of the Enterprise Architecture team at Vancity and an Architect at IBM.

Hossein Hajimirsadeghi

Date: Wednesday, January 29, 2025

Time: 3:00 - 4:00 PM PST

Location: TASC 1 9204, Burnaby campus

Talk Title: What is it like to pursue a career in Artificial Intelligence?

Abstract: Artificial Intelligence (AI) is revolutionizing industries, reshaping economies, and redefining the boundaries of technology. But what does it take to build a career in this rapidly evolving field? In this talk, we will explore the journey of pursuing a career in AI, providing insights into the diverse roles available, the skills and knowledge required, and the challenges and opportunities professionals face along the way.

Speaker Info: Hossein Hajimirsadeghi is a Principal Researcher at Borealis AI鈥檚 Vancouver lab. He earned his Ph.D. in Computer Science from 大象传媒 in 2015, under the supervision of Prof. Greg Mori. Before joining Borealis AI, he served as a Principal Member of Technical Staff at Oracle Labs in Vancouver. His research spans a wide range of AI topics, including representation learning, meta-learning, generative models, computer vision, robotics, and evolutionary optimization.

Search
Search
close