Why Artificial Intelligence Experts Call Self-Driving Cars The New “Space Race”
“Delivering the safest and most enjoyable autonomous rides pushes the limits of AI.”
In order to build the world’s most advanced, all-electric, self-driving car technology to safely connect people with the places, things, and experiences they care about, we need talented engineers from diverse backgrounds and disciplines. Two disciplines have become buzzwords in Silicon Valley: artificial intelligence (AI) and machine learning (ML).
A few months ago, Cruise hired two leaders for these disciplines: Hussein Mehanna and Tianshi Gao. Years prior, the two worked together at Facebook, only to find themselves partnering once again at Cruise.
In the below Q&A, Tianshi and Hussein share their thoughts on the AI space, what it takes to be successful, and what they hope to accomplish at Cruise.
Q&A with Cruise’s Artificial Intelligence and Machine Learning Leads
Tianshi, could you describe your relationship with Hussein? How long have you worked together?
Tianshi: I met Hussein back in 2012 when I joined the Core Machine Learning team at Facebook. We worked together and transformed the machine learning landscape for personalization applications at FB multiple times.
In both of your opinions, what mindset do you need to be successful in the AI space?
Tianshi: There are three mindsets that I think are important for the success of AI.
The first is a data mindset. AI runs on data. Embrace a data-centric view by treating data as a strategic asset, building strategies around it, being relentless about data quality, and letting data speak, write programs and help decision making.
The second is a scientific mindset. AI is mostly an empirical science. It follows the cycle of formulating hypotheses, designing and running experiments, analyzing the results to validate, invalidate, refine and inspire hypotheses, and then repeat. It requires fast exploration and being open-minded.
The last is an engineering mindset. AI is engineering. It requires infrastructure, tooling and automation to speed up the development cycle through developer efficiency and scale up AI through system efficiency.
Hussein: To be successful in AI, you want to combine skill sets in research and execution together, which requires curiosity and willingness to change your mind if proven wrong. But at the same time, delivering AI projects is hard and thus requires experience and execution discipline. Perfecting the prototype-to-production or research-to-production cycle is necessary to be successful in AI.
Knowing what it takes to succeed comes with experience. How long have you both been working in the industry? Has it changed a lot in the last few years?
Hussein: I have been working in the industry since 2003, after I did my Masters in Speech Recognition. It has changed significantly, mostly around the availability of data. In the early days, AI applications were quite limited and unreliable. Fast forward to today, we are witnessing a revolution in AI algorithms, tooling, educational materials, and most importantly, applications making their way to everyday life. We haven’t witnessed a revolution in technology like this since the Space Race.
Tianshi: I have been working on AI/ML for over a decade. In 2017, NeurIPS, a top-tier ML conference, sold out in about two weeks. In 2018, just one year later, NeurIPS sold out 1000 times faster — in less than twelve minutes. This gives a gist of how significant and fast this field has been evolving.
These changes and progress are also very comprehensive, spanning the ever-increasing data in terms of volume, diversity and richness, more powerful and efficient compute from server to edge, more advanced learning algorithms and principles that enable algorithms to develop algorithms, i.e. AutoML, better machine learning engineering with better tooling and best practices, and simply more talents and resources drawn into this field.
When all these factors combine, exponential growth happens.
Speaking of exponential growth, Cruise has grown rapidly in the last year alone. What inspired you to join Cruise?
Hussein: I wanted to help make self-driving cars a reality for the greater good. I believe self-driving technology will usher in the next revolution in tech, similar to how the Apollo program inspired generations of scientists and engineers and produced numerous technologies that we take for granted today.
In addition to making self-driving cars a reality, I was very impressed that Cruise made significant progress in the space with its brilliant team of engineers and its engineering leadership under Kyle Vogt and Mo ElShenawy. Cruise’s partnership with OEM partners like GM and Honda is also a critical advantage over other players in the market, making Cruise best positioned to deliver self-driving cars at scale.
Tianshi: I agree. I also wanted to help develop autonomous vehicles for the greater good. I joined Cruise to be part of the team that brings autonomous driving to this world, using machine learning and artificial intelligence to get us there. In a few ways, working on self-driving cars reminds me the Space Race: the problem is complex, requires applied research, utilizes the intelligence and skills of thousands of people, and when you think of the epidemic of car accidents caused by human error, there’s a sense of urgency involved.
What are your thoughts on Cruise’s vision and direction, and your role in helping us achieve our vision?
Hussein: I can’t imagine a self-driving car on the road without AI. I am looking forward to pushing the limits of AI so that we can deliver the safest and most enjoyable autonomous vehicle rides.
Tianshi: I’m excited about the AI/ML-first direction. We can easily train our AI on a number of miles much larger than a human can drive in his/her entire life. This is how we can build the world’s safest and most skilled driver. I want to leverage my experience and expertise in AI/ML to accelerate our transition to a AI/ML centric paradigm and continue to innovate to solve autonomous driving.
What do you hope to achieve while at Cruise? What do you want your legacy to be?
Hussein: Being part of what I think is the biggest engineering challenge of our generation. It feels like I joined the Apollo programs in the 50s and 60s.
Tianshi: I hope to help build the best AI team, best AI system, and best AI algorithm to enable the best driver for this world. On a personal level, I hope I can be part of the collective brain that discovers and develops the learning principles for driving autonomously.
Hussein: Self-driving cars are the next frontier for AI and AI experts. I can’t think of a more exciting application for AI!
Join Hussein and Tianshi at Cruise
If you would like to join Cruise and work with Hussein and Tianshi, take a look at our open roles. You can find their team under “AV — Perception”.
About Hussein Mehanna and Tianshi Gao
Hussein Mehanna, Head of AI at Cruise
Hussein is an expert in AI with a passion for machine learning. He has over 15 years of experience and has successfully built and led AI teams at multiple Fortune 500 companies.
Prior to Cruise, Hussein led the Cloud AI Platform organization at Google. Under his leadership, his team revamped the product line and rebuilt the organization.
Before Google, Hussein worked at Facebook where he co-founded the Applied Machine Learning group that combined applied research in machine learning and advanced platforms. He also helped democratize artificial intelligence with more than 2000 engineers using the technologies.
“My mission is to build transformative technologies by inspiring, growing, leading, and collaborating with world-class AI science and engineering teams.”
Tianshi Gao, Principal AI Scientist at Cruise
Tianshi is a machine learning expert and a published author with over a decade of experience.
Before joining Cruise, Tianshi was the lead ML scientist and engineer at Facebook. He led and scaled a team of 50+ ML engineers, scientists, and managers to increase ad matching and conversion rate through ads ranking and personalization. His team developed fundamental ML and AI technologies spanning sparse neural network architectures for personalization, distributed training algorithms, deep neural networks for content understanding, counterfactual evaluation and learning, and more.
Tianshi is also a prolific author, having published nine papers in five years. Three of his papers were selected to be presented at NIPS and ICCV, an honor received by less than 7% of all submitted papers. He has been published as first author in every top computer vision and machine learning conference, such as CVPR, ICCV, ECCV, ICML, and NIPS (now known as NeurIPS).
“I believe in scientifically and philosophically understanding the world to build a better future.”