Team update

We’ve hired⁠ more great people to help us achieve our goals⁠. Welcome, everyone!

• Dario Amodei. Dario⁠(opens in a new window) was one of the lead authors of Deep Speech 2⁠(opens in a new window), a speech system which achieved near-human performance on many speech tasks. He is also a main co-author of “Concrete Problems in AI Safety⁠(opens in a new window)”, which highlights issues related to accidents in machine learning systems. Prior to OpenAI, he worked at Google Brain.

• Filip Wolski. Filip’s⁠(opens in a new window) recent background is in “practical” modeling, having spent the last few years working in the high-frequency trading space. In the past he enjoyed problem-solving in programming competitions, and won the IOI⁠(opens in a new window) and ACM ICPC⁠(opens in a new window).

• Jack Clark. Jack⁠(opens in a new window) has spent the past few years writing about artificial intelligence and distributed systems, most recently at Bloomberg and BusinessWeek. His articles have covered technologies like memory networks⁠(opens in a new window), image generation⁠(opens in a new window), and reinforcement learning for robots⁠(opens in a new window), and issues like diversity within AI⁠(opens in a new window). As our Strategy and Communications Director, he will help with community outreach, policy, communications, and strategy.

• Scott Gray. Scott⁠(opens in a new window) was previously an engineer at Nervana Systems⁠(opens in a new window) where he focused on optimizing the performance of deep networks on GPUs. His assembly-level optimizations⁠(opens in a new window) for dense linear algebra and convolution remain the fastest available. When not writing software⁠(opens in a new window) he’s usually spending his time reading up on the latest research in neuroscience and related fields.

• Zain Shah. Zain⁠(opens in a new window) previously led deep learning efforts to build a collaborative human-machine intelligence system at Clara Labs⁠(opens in a new window). He’s also worked on speech synthesis⁠(opens in a new window) and computational neuroscience⁠(opens in a new window), built Mosaic⁠(opens in a new window), and founded a mobile behavioral analytics company⁠(opens in a new window). He most recently built a GIF search engine⁠(opens in a new window) using deep multimodal embeddings.

We’re also pumped to be working with the following people for a more limited period of time:

• Catherine Olsson. Catherine⁠(opens in a new window) built OpenAI Gym’s REST API⁠(opens in a new window), which has already attracted users in Lua⁠(opens in a new window), C++⁠(opens in a new window), Java⁠(opens in a new window), and Rust⁠(opens in a new window). She graduated with a perfect GPA in CS and Brain & Cognitive Science from MIT, and has extensive research experience in computational neuroscience⁠(opens in a new window) and psychology⁠(opens in a new window). Catherine has taught programming and applied math⁠(opens in a new window) for six years, including outreach to women and underrepresented minorities.

• Harri Edwards. Harri⁠(opens in a new window) is a PhD student at the University of Edinburgh, where he is researching models that can quickly adapt to new situations by learning to represent datasets⁠(opens in a new window).

• Igor Mordatch. Igor⁠(opens in a new window) is interested in optimal control, machine learning, and their applications to robotics, biomechanics, and neuroscience. His PhD was in automated discovery and learning of complex movement behaviors⁠(opens in a new window). He will join the faculty at CMU⁠(opens in a new window) in September 2017.

• Taco Cohen. Taco⁠(opens in a new window) is a PhD student working on applied and theoretical problems in representation learning. Most recently he invented group equivariant convolutional neural networks (G-CNNs⁠(opens in a new window)), a generalization of CNNs that improves the statistical efficiency of these models by exploiting symmetries.

• Tambet Matiisen. Tambet⁠(opens in a new window) is a PhD student from University of Tartu, Estonia. He previously worked as a software engineer and founded his own startup⁠(opens in a new window). His recent projects range from making deep reinforcement learning agents cooperate⁠(opens in a new window) to predicting a rat’s location from its brain activity. He also wrote an accessible introduction to deep Q-learning⁠(opens in a new window).