Joel Pobar is a technologist and engineering leader known for his work on software platforms and artificial intelligence projects. He has held senior roles at major technology companies, including leading infrastructure and engineering at Anthropic and serving as a senior engineering organizational leader at Meta, where he worked on AI initiatives including PyTorch and was associated with the Superintelligence team. ^[1]

Education

Joel Pobal completed a Bachelor of Information Technology with Honours in Computer Science at the Queensland University of Technology (QUT) between 1999 and 2001. ^{[3] [4]}

Career

Joel Pobar began his professional career in Brisbane as a Systems and Network Administrator at Quicknet Internet Provider, where he was responsible for maintaining IT infrastructure across multiple platforms, including Windows, Linux, FreeBSD, and Cisco-based systems, between 1999 and 2003. Concurrently, he worked as a Research Assistant at the Queensland University of Technology (QUT), participating in projects related to programming languages, distributed computing, and software adaptation frameworks. In 2002, he completed an internship at Sun Microsystems in Menlo Park, contributing to the development of a code partitioning system for an internal Verilog compiler targeting ASIC-based execution environments.

In 2003, Pobar joined Microsoft as a Program Manager on the Common Language Runtime (CLR) team. His responsibilities involved work on several components of the .NET framework, including reflection, code generation, and delegate mechanisms, as well as contributions to the Shared Source CLI (SSCLI). He later collaborated with the Microsoft Solutions Development Centre (SDC) as a contractor, participating in technical architecture and engineering efforts across multiple large-scale software initiatives. In 2010, he served as a contracted engineer at Macquarie Bank in Sydney, where he helped develop trading infrastructure designed for high-throughput and low-latency operations.

Pobar moved to Facebook (now Meta) in 2012, where he held successive engineering leadership roles for over a decade. Early in his tenure, he managed teams responsible for core software infrastructure projects such as HHVM (a just-in-time compiler for PHP), the Hack programming language, and the Flow static type checker for JavaScript. Between 2015 and 2019, he directed initiatives focused on performance infrastructure across Facebook’s applications, overseeing optimization efforts for mobile and backend systems. From 2019 to 2023, he worked with AI and machine learning infrastructure teams, contributing to internal platforms used in model deployment, orchestration, and large-scale training workflows, including work involving PyTorch and FBLearner.

In 2023, Pobar joined Anthropic, where he is part of the technical staff focused on engineering infrastructure for inference systems supporting large language models. In 2024, he became the first Venture Partner at TEN13, a venture capital firm with investment interests in artificial intelligence. His role involves advising on technical aspects of AI-related opportunities and contributing to the firm’s investment strategy.

Pobar has also been involved in technical publishing and conference presentations. His contributions include articles for MSDN Magazine and talks at industry events such as the GOTO Conference in Aarhus, where he spoke on the use of functional programming techniques in applied computing contexts. ^{[2] [1] [3] [4] [5]}