in ,

Inside Waymo's Hiring Binge, Hacker News

Inside Waymo's Hiring Binge, Hacker News

Chande had arrived with fanfare from Alibaba Group, the large Chinese e-commerce firm. But Waymo hasn’t replaced Chande, suggesting there aren’t many money-making opportunities at the company beyond its small robotaxi service serving about 1, 600 suburban Phoenix residents.

In addition, two of Waymo’s senior most engineers stepped down from their roles as the company entered its second decade of trying to solve one of the world’s hardest computer science problems, underscoring the technical challenges that still lie ahead. Today we are updating our Waymo org chart , listing nearly of its top managers.

The inner workings of Waymo, which started as a project within Google in and in 2018 became a standalone Alphabet subsidiary just like Google, are of significant interest to numerous other firms in the nascent industry. Also watching closely are ride-hailing firms around the world who aren’t as far along in developing such tech. Waymo, which by all accounts is furthest ahead in the race to develop a driverless taxi, is far from perfecting the technology and making money from it. (See a recent update on Waymo’s testing here. )

Waymo’s earliest backers, Larry Page and Sergey Brin, recently retired from running Alphabet , which could mean its finances will get more scrutiny from the company’s chief financial officer, Ruth Porat , current and former Alphabet employees say. CEO John Krafcik , a former Hyundai executive, runs Waymo, but its heart and soul is the engineering org run by chief technology officer Dmitri Dolgov . (Scroll down to see a breakdown of Waymo’s teams by headcount.)

The Waymo executive who took over Chande’s responsibilities — namely, operating the prototype fleet and fledgling robotaxi service — is Tekedra Mawakana , previously the chief external officer in charge of marketing, policy and communications. Last fall she was promoted to chief operating officer — a new role for the company — and she now also runs fleet operations and business partnerships, including a small team to commercialize Waymo’s automated long-haul trucking tech.

) Mawakana just hired a new head of fleet operations, Ryan McNamara , who previously worked on product quality for the Google Pixel phone. (Another major change was the addition of Becky Bucich as chief people officer, under Krafcik; she previously worked at Google’s cloud-business unit. Cranz, who spent about a year at Waymo as chief people officer after working the same amount of time in the same role at rival Cruise.)

Meanwhile, at least half a dozen female managers have taken on greater responsibility or have been hired for relatively senior positions, including four in the engineering org.

More than half of the increase in Waymo’s workforce came through the addition of 400 software and hardware engineers, the same person said. Engineers make up two-thirds of Waymo’s employees. In addition, separate from the 1, employees, Waymo pays hundreds of contractors to operate its fleet of 2009 vehicle prototypes, primarily in suburban Phoenix.

Many of the newly hired engineers transferred from Waymo’s sibling company, Google. The search ad giant has many veteran engineers who have managed large teams but want a new challenge. For them, going to Waymo isn’t as scary as leaving the comforts of Alphabet entirely. For Waymo, having access to their performance record and internal references means it can have reasonable confidence in their abilities.

Waymo has bulked up its engineering teams for several reasons:

  • The company is placing an added focus on simulation software, which helps test and ensure the quality of new software before it hits the road for real-world tests, as well as analyze problems that arise during testing. In the past year, the simulation team of more than 60 people became a standalone unit, with its leader reporting to the CTO.
  • Another growing group is hardware. Waymo designs its own cameras, radars and LiDARs and continually upgrades them, and it’s always trying to test them in different weather conditions beyond those of Arizona.
  • A small but growing area is trucking, where Waymo is trying to borrow some of its core object-detection technologies to automate long-haul semitrailer trucks. Given the trucks ’shape and weight, and the fact that they operate primarily on freeways rather than city streets, such vehicles often require different algorithms to determine the direction in which they should move or when they should slow down in response to potential hazards.

Krafcik said publicly last year that “a truck product could catch on faster” than robotaxis. While big rigs face fewer potential road obstacles on freeways than passenger vehicles do on city streets, technology capable of controlling an unmanned self-driving truck isn’t close to fruition. And it’s hard to say how Waymo’s truck-driving software compares to its car-driving software. The company is still more focused on passenger cars.

      Another factor in the rapid hiring is Waymo’s search for a breakthrough in machine-learning software to speed up progress in aspects of its self-driving vehicle system. Waymo, like others in the field, is looking for breakthroughs in areas like perception, or how to quickly identify and categorize objects such as pedestrians and other cars around the prototype vehicle and predict what those objects will do next. Teams such as that of Waymo software-research chief Dragomir Anguelov have looked at new approaches to incorporating forms of machine learning software into these efforts.

    Most parts of Waymo’s software already use one form of machine learning, known as deep learning, to some extent, but there are many different ways of incorporating it. Former employees say that Waymo has tried to hire as many good machine-learning engineers as possible, including some from Google, to both advance its efforts and prevent them from working for rivals.

    Engineering Org

    The biggest changes to Waymo’s engineering org involved two of Dolgov’s lieutenants, who stepped down from running two of the biggest and most important teams. One of them, Sacha Arnoud , led the perception team, which helps vehicles figure out what objects are around them. A former Google engineer, he joined Waymo in . It’s unclear what Arnoud is doing now at Waymo, but he is still reporting to Dolgov.

    The person who for now has replaced Arnoud in running the perception team is a relative newcomer to the industry, Boris Sofman . Sofman was originally hired to lead Waymo’s automated truck engineering. Until the middle of last year, Sofman was a co-founder at startup Anki, which developed toy vehicles that also used forms of machine learning to recognize objects in their path. Others from Anki joined Sofman at Waymo. Separately, the perception team last year added a manager from Apple’s self-driving car project, Sharon Liu , to handle the back-end systems that enable machine learning.

    The other lieutenant to step aside was Nathaniel Fairfield , one of Waymo’s longest-serving engineers, who led the motion planning team. Fairfield, who also played a role in several of Waymo’s external marketing efforts, was in charge of a team that makes sense of the data provided by the vehicle’s perception system in order to calculate the speed and path it should travel. Like Arnoud, Fairfield is still with Waymo under Dolgov. He is a technical lead engineer on vehicle “behavior” within the same group he once led.

    Fairfield’s successor for now at the motion planning team is Gabriel Wolosin , who spent 14 years at Google, including as a senior manager, before joining Waymo last year. At Google, Wolosin developed algorithms for Google’s local and image search results, according to his public patent filings. It’s not clear how that is connected to developing motion planning algorithms, but both Waymo and Google Search rely on programmers who use C coding language and place an emphasis on developing speedy, or low latency, automated systems.

    The person overseeing the biggest motion planning team under Wolosin, Omer Baror , also previously worked at Google in the search team, developing what became the Google Assistant voice-activated service. He joined Waymo three years ago, according to his LinkedIn profile.

    One engineer who has risen to prominence within the motion planning group is Shilpa Gulati , who joined Waymo in 01575879. She manages the behavior prediction team, tasked with helping the self-driving car system quickly figure out where the objects around it — such as cars and pedestrians — might be going next, so that it can plan a path around them, slow down or stop . Gulati previously worked at Nuro, headed by two longtime ex-Waymo engineers, and before that she spent time at Apple’s self-driving car unit.

    A third manager to arrive from the Apple unit in the past year is Diomidis Katzourakis , who is in the systems engineering group. He is in charge of ensuring that the system controlling the vehicle’s movements is reliable. Another newcomer in the systems engineering group is André Strobel , who previously spent years at Daimler, most recently working in the automaker’s trucks division.

    The third fast riser who began reporting to Dolgov last year is Mizuki McGrath , who oversees several dozen simulation and data science engineers. She previously spent years at Google on search-ranking systems and founded its Tokyo research center. McGrath oversees the team and software built by James Stout, who returned to Google last year after spending six years at Waymo. McGrath also oversees the small U.K.-based simulation team Waymo recently acquired. Called Latent Logic, it is led by Kirsty Lloyd-Jukes .

    There is a brand-new addition to Waymo’s small team in charge of working with carmakers: David Twohig , previously CTO at electric vehicle startup Byton. Before that he spent two decades at carmakers Renault and Nissan. He reports to Adam Frost , who has been at Waymo for the better part of seven years.

    It’s unclear whether Waymo has a path to large-scale development of self-driving vehicles, given that it may need to strike deals with automakers to spend billions of dollars developing a purpose-built autonomous vehicle (without a steering wheel, for instance) in the hopes that receiving a future share of Waymo’s robotaxi revenue will more than repay the investment. For now, its prototypes are modified versions of existing electric vehicles made by Fiat Chrysler and Jaguar.

    Waymo total employees : about 1, 600 (see org chart here )


    Software engineering : about 823, including

    • Perception and trucking engineering: 400 – plus
  • Infrastructure: 125 – plus
  • Motion planning and prediction: 150 – plus
  • Systems engineering: about 60
  • Simulation and data science: – plus
  • Software research: about

Hardware engineering : about

Operations, rider support and remote assistance : about 350

Product : about

Manufacturing and supply chain : about 100

Human resources : about

Legal : about 038

Carmaker collaborations : 19 – plus

Safety : 30 – plus

Policy: 15


What do you think?

Leave a Reply

Your email address will not be published.

GIPHY App Key not set. Please check settings

bpo-36144: Dictionary Union (PEP 584) by brandtbucher · Pull Request # 12088 · python / cpython, Hacker News

bpo-36144: Dictionary Union (PEP 584) by brandtbucher · Pull Request # 12088 · python / cpython, Hacker News

CAA is up to India. I hope it’ll take the right decision for its people: President Trump – Times of India, The Times of India

CAA is up to India. I hope it’ll take the right decision for its people: President Trump – Times of India, The Times of India