Meet Charlotte, a PM on Lyft’s Autonomous Vehicles Team
Autonomous Vehicles (AV) is a relatively small and niche space. What are your previous experiences that brought you to AV?
Before getting into the autonomous vehicles space, I had worked in data science for 5 years as a technical individual contributor. Three years ago, I started specializing in data-related products for autonomous vehicles and decided to stay in the AV space since then. I am now the Product Manager for Mapping at Lyft’s autonomous vehicles team (Lyft Level 5), supporting engineers from Palo Alto, Munich, and London. Though mapping is unrelated to data science, I can still leverage my data skills to design metrics and measure product performance.
Having a data science background is not uncommon in the AV field. You would be surprised by the rich diversity of job skills that the AV space requires. Other than software development and machine learning, specialities such as aviation, UX, fleet operations, and infrastructure, are also very common skills seen across our teams.
So you’ve had prior AV experience! What keeps you motivated and wanting to stay working in AV?
AV is probably the only space where you can accomplish three things via one job: taking on the most cutting-edge technical challenge (for professional development), improving efficiency for getting from point A to point B (building a valuable product), and saving people’s lives through safer transportation (for a mission-driven profession). You rarely find all three in other industries. Many other technical products either cannot promise a commercial launch or do not relate as closely to our daily lives.
Working as a Product Manager for AVs is intrinsically interesting. You have to accept the fact that the product you build is as hard as aerospace engineering, but somehow the “rocket” needs to be launched to the public in a safe, timely manner, which entails extremely careful product roadmapping and prioritization. You would know that the “rocket” is not there yet when your autonomous vehicles still struggle with making left turns. You would know that your rocket cannot “land on the moon” when your driving trajectories still require some fine-tuning. The impulse to overcome these challenges is what keeps me up at night and wakes me up in the morning, everyday.
What made you choose to work on AV at Lyft specifically, as opposed to other AV companies out there?
Lyft is well-positioned to launch autonomous vehicles at scale because it enjoys the luxury of its large fleet from Lyft RideShare. There are many ways through which AVs can benefit from human drivers. Mapping is a notable example. Maps are consumed by both human drivers and autonomous vehicles but in different ways. Human drivers need general navigation from maps, whereas autonomous vehicles need high-precision road features from maps. At Lyft, teams from both Level5 and RideShare collaborate actively to maximize the value from one common map. Lyft provides the full freedom for teams to explore how humans and robots can cooperate.
One other reason why I chose to work at Lyft: culture. Building an AV is a marathon without a clear course map. That’s why company culture plays a critical role in AV development. Lyft’s culture is ‘quality over speed’. Quality is the foundation for safe operation of AVs. We also value collaboration over individualism. Building AVs requires teamwork across software, hardware, operations, and many other functions. At Level 5, decisions are typically made with all stakeholders in the room so that all considerations can be taken into account. Ultimately, it’s the culture, rather than the tech, that brings talent together to deliver reliable autonomous vehicles.
Thinking back to your experiences before AV — how is being an AV PM different from other PM experiences or other AI related products?
Autonomous vehicle products are fundamentally different from traditional products in many ways. Firstly, the AV product life cycle is much longer, sometimes a year, sometimes even three to five years, because every product feature has to go through in-depth research, pioneering prototyping, rigorous testing, and thorough validation. Each phase can take months or even years. Adding a brand new layer to maps, for example, can take a long time to implement. Product Requirement Documents (PRDs) can become obsolete every time product goals shift. More likely than not, teams are unable to deliver any tangible products in a long period of time due to technical difficulties. Successful PMs need to break down large products into small milestones so that the incremental improvements can be celebrated along the way.
Some incremental improvements can be iterated rather quickly. One code change in the morning may fix a disengagement during a test drive in the afternoon. Nevertheless, teams need to be constantly reminded that not all technical solutions are worth exploring. PMs help engineers focus on long-term solutions without compromising short-term deliverables. Some quick wins could simply be red herrings. Patience is a prerequisite to becoming an accountable PM in the AV space.
Secondly, there are no prior product best practices that you can borrow from the industry. Unlike conventional consumer products, no one has successfully launched level-4 autonomous vehicles to the public at scale. Our Open Platform team has enabled Lyft to become the only company that has delivered a large volume of autonomous rides with with a safety driver at the wheel. Being the first comes with challenges. As a PM, you get to define your own standard, even when it is wrong. This is what makes being an AV PM such an exciting job. Trial and error is common in developing product roadmaps, but acknowledging and embracing failures will make your team and products stronger in the long run. Take mapping as an example. There is no industry standard for creating high-definition maps for autonomous vehicles. Therefore, we created one that we believe can set the team up for success.
Besides trial and error, there are many other ways product managers can adopt to build product roadmaps. Autonomous vehicle PMs also learn their skills through attending industry conferences, talking to thought leaders in academia, and reading the latest research papers. To foster mutual learning, PMs are also responsible for sharing internal knowledge with external stakeholders. Lyft Level 5 just recently launched a valuable dataset to the public with the hope that the community can work together to design best practices for autonomous vehicles.
Lastly, you need to manage your customers when there are no real customers yet. In the traditional product space, products have clearly defined user groups from which product managers can receive instant feedback. In the autonomous vehicles space, however, your customers are your vehicles and the engineers who build them, since you do not have real riders yet. Establishing a smooth internal feedback loop is critical while your product remains in the R&D phase.
Within the feedback loop, both upstream and downstream stakeholders are equally important. Unlike developing traditional online products, to deliver an autonomous vehicle product, even one small task can still involve a handful of engineers across multiple teams. Way too often, some stakeholders’ feedback is neglected, which leads to product failures at a later stage. Managing stakeholder relationships is an art for AV PMs. You succeed or you fail together. PMs always have to revisit their mental stakeholder relationship map to ensure all feedback is taken into consideration.
Do you think that PMs who work on AVs need to have a technical background? It’s obviously a very technical problem to be working on.
It may sound surprising, but Lyft’s hiring committee has made it crystal clear that prior AV experience should not be a hard requirement. Lyft believes that new hires can gain deep knowledge in autonomous vehicles once they join the team. Lyft’s collaborative culture makes employee onboarding extremely easy. That said, PMs are expected to speak the technical language with their engineers. They live and breathe by how well they understand the technology and should never stop learning new things. This is largely how AV PMs turn themselves into high-performing employees over time.
Since it’s a relatively new industry, there isn’t a lot out there on what it takes to be a good AV PM. What can candidates do to set themselves up well to be in the AV space?
Although prior AV experience is not a hard requirement, some high-level knowledge about how AVs work is still preferred. Some general ideas about AVs can help candidates understand the various products in the AV space and anticipate how to manage their future stakeholders with diverse technical backgrounds.
Most candidates learn about the industry by following some recent news about AVs, but be mindful that most news articles are written by folks who do not necessarily work in the AV space. Try to think deeper than what you see at the surface level. For example, many articles talk about how frequently different autonomous vehicles disengage during public road tests, but the fact that disengagement is a shallow metric has already been acknowledged by many AV experts. PM candidates are encouraged to build new metrics to accurately measure AV performance.
In addition to what is written in job descriptions, candidates are also expected to master product prioritization skills. Due to lengthy product life cycles, teams constantly look for guidance from their PMs with regard to what to build next week as opposed to what to build next year. A team is healthy when its engineers rarely struggle with prioritization.
You could be next
If you want the experience of creating a product from 0 to 1, then joining the AV space would definitely be a wise career choice. Product management for AVs is fundamentally different from traditional product management. You get to contribute to industry best practices since you are one of the first few players in the space. Lyft provides the perfect culture and platform for teams to develop best practices and encourages every employee to become an AV thought leader. You could be next.
Shout out to our Mapping team at Level 5, without whom this blog post wouldn’t have existed. Our Mapping engineers are the heroes behind the scenes. AVs can drive with peace of mind thanks to high quality maps. Please come join us!
Lyft AV is hiring! Our Mapping team is hiring. Our Product team is also hiring! Check out lyft.com/careers. We have open roles across Palo Alto, London, and Munich. Read more about Lyft Level 5 here.
