← The Digital Executive
The Digital Executive
Ani Mishra on Scaling Logistics and Engineering Leadership at DoorDash | Ep 1087
00:13:52
Loading summary
Transcript14 lines
- [00:00]
A
Foreign.
- [00:09]
B
Welcome to Coruscant Technologies, home of the Digital Executive Podcast. Welcome to the Digital Executive. Today's guest is Ani Mishra. Ani Mishra is a Seattle based software engineering leader with a proven track record of building consumer facing products that achieve widespread adoption. As the head of DoorDash's new Verticals Logistics Engineering organization, ANI is responsible for the 24 by 7 operational excellence of a large scale system spanning grocery, convenience, retail and alcohol. His team's critical work ensures the seamless and optimal matching of orders, dashers and stores directly influencing key services in DoorDash's consumer and Dasher apps. Well, good afternoon Ani. Welcome to the show.
- [00:50]
A
Hey Brian, thank you so much for having me on the show.
- [00:53]
B
Absolutely my friend. I appreciate it and making the time. No, you're in Seattle, Washington today. I'm in Kansas City, so there is a two hour difference. I appreciate you making the time. Ani, I'm going to jump right into your first question. You've led engineering from some of DoorDash's most complex logistics systems. How do you approach designing a platform that operates seamlessly across grocery, convenience, retail and alcohol, each with its own set of challenges?
- [01:19]
A
That's a good question, Brian. And building a system that solves customer problems for multiple verticals and is no EV tab like. My philosophy for building systems that can solve many customer problems in a scalable way is start with one of the categories of customers. So define the customer problem really well, understand the problem well and build a solution that solves the problem for one category of customers and figure out how to build a product that works for the customers. Get it out as soon as possible. And once you have figured out what these should be built to solve the problem for one set of customers, that is when you start to like think about how do I scale this solution. So one thing that I keep in mind is even if I'm building a solution for one set of customers, like what would I need to build for it to work for other type of customers or other kind of problems. So keep that in mind while building the system or the platform initially and optimize for speed initially. Once you've figured out what you're building what your customers want, that is the time to think more about scaling it and building a more generic platform. How I like to do this is just figure out the pilot use cases for the product you're building and get it work for them for one set of customers and then continuously iterate and keep generalizing the solution and until you reach a Point where your solution works for multiple set of customers and multiple verticals. So the solution kind of evolves from solving one specific problem for one set of customers to solving many problems for a lot of customers. So how I approach start simple, start small, go to market fast, learn from there, iterate and then think about building the platform and generalize your solution to work for a larger of customers.
- [03:01]
B
Amazing. Thank you for breaking that down. Obviously there's a lot that goes into what you do in your job. I liked how you tackle solving multiple problems across all these verticals as we discussed. But focusing on one set of customers, define the customer problem. How do you build that product for that one customer, optimize for speed and then scale again? You can do this. It's kind of like a rinse in repeat. And I appreciate your experience and how you broke that down so easily for us here in our audience. Bonnie, the next question I have for you. As a leader in such a fast paced engineering environment, how do you build and maintain high performing teams that can both innovate quickly and operate reliably at scale?
- [03:40]
A
Brian, so that is a good question as well. So having a team that can actually build system tag with and there's a lot that goes into building a team that is excited about solving problems, challenging problems at scale, not just solving customer problems, but also solving large scale systems problems. My approach to this is, you know, first we need the right people with the right skillset in the room. So how do we hire top talent? How do we attract, you know, world class engineers to work with us on solving these problems? You know, what I look for is engineers who actually are passionate about solving customer problems. Obviously computer science background and large scale systems background is also necessary. But people who are really excited about solving customer problems, like applying the skills in solving these hard customer problems is really critical. Let's say you figured out how to hire top talent. Understanding motivation of the people in the team is also very important. Some people are just excited by solving hard technical problems, but there are other people who are excited by leading other people. Some people like to manage projects. Some people like to, you know, get to the next level in their careers. I think understanding motivations for the people in my team is, is also very critical to make sure that folks stay motivated. And also like for me as a leader, it's very important to have enough charter or enough scope so that I can set everybody in my team in their growth path. So it's very important for me to continuously keep looking one year, two year out and thinking about the problems that I want to solve in the future and start to set the stage for solving these problems, get the prototypes out, create enough traction within my org to be able to start solving those problems. Finally, like, one of the most important things I think for like high performing teams and you know, working with high performing engineers is giving people focus. It's very important for, you know, me to set very clear focus and very clear performance space for the people in my team. And making sure they have autonomy to make decisions of their own is really critical for people to do their best work. Obviously another thing that is very critical is how do you build systems which are really reliable, which requires a very disciplined approach to building systems. For me, like I said that as a goal for everybody in my team that they are actually responsible for the reliability of the system as well, whether the system needs four nines, five nines. It is for them to decide based on the criticality of the flow. And just building the product is not enough. The product should also be reliable for the customers. That is kind of like how I think about building high performing teams in fast paced cultures operating at really large scale.
- [06:07]
B
Thank you, that's awesome. Broke a lot down there for being a successful team. Obviously having that good team to solve these complex problems and innovate is so critical these days. If I could just highlight a couple of things, things that you mentioned, obviously hiring that top talent, find out what motivates them, you know, are they passionate about solving customer problems, defining a clear path and support for your team, foster an environment and culture with clear goals and give them that autonomy. And lastly, I think it's important is reliability is key, but everybody shares in that responsibility and I like how you share that message as a team. Bonnie, with your early experience at startups like Phoenix, P2P and Mobifi, how did these roles shape your approach to innovation and leadership at a much larger company like Doordash?
- [06:53]
A
My background working for really small startups, so I was founding engineer for one of these startups and I was really like first 10 employees for the other startup. And I think I really learned important lessons working for really small companies early in my career that helped me grow my career in future and like becoming, you know, a leader at Doordash. And I think what really helped me was that I really embraced my time at the startups and really embraced what it takes to like, you know, write the first line of code for building a product where we don't know where we are going and then going from there to a stage where the product is mature and there is people that depend on that product for their everyday life. So, you know, learning the transition from building an early product to going to a mature product and what it takes to do that, I think that is one of the most important lessons that I learned working for really small early startups, particularly at Phoenix, which is a Chicago based company that works on real time video streaming at large scale. I learned how to build large scale systems that scale. And you know, I really remember like one time I thought this system is so complex, how do I understand it well enough to be able to, you know, build a system on top of it and extend the capabilities? And you know, I signed up to be on call for a month so that I can understand the internals of the system. And I still remember those were some of the most formative days for me in understanding, you know, how to build large scale systems. And you know, if I reflect at my time at Mobify, I think I really learned how to lead teams and how to build inclusive cultures there. And even today I apply some of those learnings in enhancing my team and building a very inclusive culture here at Doordash. I definitely think those are some of the most formative days in my career and I recommend everybody to, you know, try a stint at an early startup at some point in your career because those are, you know, the really important and you know, the most really critical for me to like learn and you know, execute in future.
- [08:46]
B
Absolutely, I appreciate that and I hear a lot of that from founders and people that were like yourself, founding engineer as an example, right? Startups, you wear so many hats. When you're at a small company, you're asked to do a lot more, but it gives you that appreciation of what goes into a startup. You know, wearing many hats, you've learned how to build a product early, right? And then learn how to evolve into that mature product. And you really learned how to scale products, which I think is pretty cool. So thank you for sharing. Ani, last question of the day. What excites you most about the future of logistics technology? Whether it's automation, AI or emerging delivery modes, where and where do you see the biggest opportunity?
- [09:25]
A
Brian, this is a very timely question. With the, you know, with the emergence of large language models in the last few years now a lot more is possible than what it was five years ago. And I think while the use of AI is not novel in logistics like we have seen research and work done on autonomous vehicles and drones and, you know, food making robots, automation for many years and you know, AI has been playing an important role in advancement in those areas. But the emergence of large language models have made a lot more possible now because now everybody has access to word knowledge. So it has really enhanced the capabilities of all the actors in logistics. And particularly like in last mile delivery and on demand delivery space. I think large language models are already helping associates in their store that do the shopping on your behalf when you place an order. It is helping them find items in their store very easily and it is empowering them to be able to easily discover some of the most obscure items that they might not know how to find in a store. And not only it reduces the defects because people are able to associates in the store are able to find more items, it's also making them more efficient. Now they can find the fastest route to shop all the items that the customer wants. It is definitely helping reduce defects as well as making on demand delivery even faster. And in addition to that, I've also noticed in industry that it has dramatically improved the customer experience. Now a large engine model can understand the intent of the customer when they're placing an order, let's say on Thanksgiving. Your ingredients seem like you're going to prepare Thanksgiving dinner for a family. AI and large image models can understand that, hey, this person wants to prepare a dinner for Thanksgiving for their family. Seems like they're missing one of the key ingredients and they can nudge the customer. So it's also an assistant for people who are placing the orders on delivery apps and making it easier for them convey their intent more clearly. And Finally, I think LLMs are also playing an important role in safety for workers who are transporting goods for last mile delivery. And one important use case I like to think about is that a large engine model can easily understand free text information about weather, about traffic, and it can automatically adjust. It can automatically shut down a market and get it back on really fast based on the real time conditions, thus ensuring that everybody who's doing the transportation of goods, drivers, statues are really safe. So there's really a lot of applications of LLMs in logistics and we are just starting to see it all happened. And I think in future we'll discover a lot more use cases and a lot more applications, not only making these deliveries faster, but with lesser defects and safer for drivers. I would like to also add that while LLMs are new, the AI applications in logistics are not novel. That has been happening for a while. We'll continue to see autonomous vehicles becoming more and more mainstream already today in San Francisco and Phoenix in Collarwrite on autonomous vehicle Los Angeles. You can get a delivery from autonomous vehicles. That will continue to happen. It will accelerate with the help of LLMs. Another classic problem that AI solves in logistics is predicting demand. And that helps logistics companies to, you know, to make sure they have capacity to fulfill that demand. And even in that area, we'll see more and more innovation with LLMs and acceleration of all of these use cases and their solutions in future. You know, really exciting time to be in logistics, applying AI and LLMs to solve problems. And I think we'll see a lot of exciting advancements in this field in the next few years.
- [12:54]
B
That's amazing. Love hearing this type of stuff. It gets me really excited. We talk a lot about emerging tech here on the podcast, but you know, just a few things that I like to highlight you covered. With the advancement of LLMs today, there's so much more that's possible. As you mentioned, we've got these autonomous vehicles, you know, robotic process automation, drones and other types of robotics. This technology has really enhanced the logistic of products from order to delivery. As you mentioned, with improved efficiency, customer experience, you can go as far as reading customer sentiment, customer intent, which is really awesome. And the last thing you talked about is that free text information, you know, it will drastically improve logistic process while it's in process. Right. It's real time decision making, which I think is phenomenal. So I appreciate that. Ani, it was such a pleasure having you on today and I look forward to speaking with you real soon.
- [13:42]
A
Likewise. Thank you, Brian.
- [13:43]
B
Bye for now.