Nitro: Erran Berger, Q&A with Alexy Khrabrov on Engineering Culture

Nitro: Erran Berger, Q&A with Alexy Khrabrov on Engineering Culture

Recording: Nitro: Erran Berger, Q&A with Alexy Khrabrov on Engineering Culture

hello everybody I'm Alexi kro a chief scientist at Nitro and this is a nitro tech talk series and here with us I have Aron Berger senior director of engineering at LinkedIn uh and he's going to talk about uh engineering culture today he has some series of amazing posts uh on link about building engineering culture and we're engineering organization uh growing to uh to web scale and scaling so we're very interested so I'm going to ask Aran about uh his experience and first can you tell us a little bit more about your background and uh your career at LinkedIn sure so uh I've been at LinkedIn almost six years now which makes me very much an Oldtimer um started LinkedIn as an individual contributor software engineer working on profile uh this was back in 2009 right around when LinkedIn was really hitting its its its big hockey stick growth moment and uh being responsible for what is effectively the core uh uh one of the core pieces of infrastructure at LinkedIn which is you know our our uh databases storing our profile data as well as the services that expose uh that data uh we're having scaling problems and so spent the first 12 or 18 months really uh firefighting and bailing water and and uh and doing whatever we could to to uh scale up that system stabilized it and and really were able to get it to a point where uh we weren't having issues every Monday and Tuesday morning at Peak traffic like we were when I first started um and then and then once we had really stabilized uh the infrastructure we're able to Pivot and start thinking about you know what we really wanted to the the pro profile to be at LinkedIn and developed a bunch of features both that targeting you know uh and helping our members to show the best of their professional selves online uh as well as helping us you know uh um help helping them fill out their profiles which you know profile data is what powers every every part of LinkedIn experience um after that was given the opportunity to to take on a larger role at LinkedIn and Lead what we called our our core engineering team things basically our core products things like ident all the all the profile related products which is responsible for but growth and uh internationalization and the member-to-member connection ecosystem um as well as a few other products and uh that was obviously very interesting working on basically the the heart of what uh the LinkedIn product Suite is and then more most recently um decided I want to transition into a slightly less mature leading a set of products that were slightly less mature with their technology and the product strategies we still you know in their early stages uh which is how I landed my my current role uh heading up engineering for Content products things like pulse and slides share and the publishing platform and sharing of content on LinkedIn really trying to uh a set of products aimed at giving professionals the knowledge that they need uh to keep up to date what what's happening in their industry and and in their company um and to really help give them the knowledge that they need to solve the and to do the work that they are doing every day in their jobs yeah I must say you I already see a lot of this uh products surfacing you know pulse and and the SlideShare so that's I think there's a huge potential because a lot of people share you know the presentations so it's it's it's I would say this is a really amazing career right because LinkedIn is a giant organization I worked in know top prospect uh uh social hiring recruiting startup and we dealt with LinkedIn data and uh uh it was very interesting to see you know uh LinkedIn already operated as a big company with was three years ago so and and I wonder how do you um how do you create change it's such scale right you know there is this book about ABM elephants can dance can you make this elephant dance how do you actually accomplish something in such a big organization it's a really good question um the the reality is I mean like I I would be lying if I said that as linkedin's grown from the 400 people it was when I started to the nearest you know 6,500 7,000 I don't know what the number is now that that um the the effort required to really make you know significant change in impact on the businesses harder that said I think that uh ultimately making change is something that requires people who want to be champions for that change so um I'll give you an example we um for a very very long time we had uh you know in our data center we probably have something on the order of 300 350 Services the power for you know the LinkedIn experience and we had accumulated an enormous amount of technical debt around our apis they were not uh we were using spring RPC so it was not language agnostic you know um it was actually like the framework itself was pretty nasty um it was uh really hard to make backwards and comp uh compatible changes and so a set of Engineers decided you know what like this has to stop like we actually need to re ride all of our apis uh we need to build a new framework um that's going to help us model our our uh domains much easier and expose those domain models to to clients and like they they just took the mantle and they championed it and they talked to the technology executive team and and you know really got byy in you know for for this massive effort that's now taken us 18 months to accomplish if you imagine like rolling out domain model and and API changes across 350 services and data center is non-trivial and yet you know a handful of Engineers decided that this was important enough for the company's future to do so ultimately I I I believe that no matter H how large the company is if you have Engineers who are passionate uh about doing things the right way and about change and you have a a leadership team at all levels that supports that kind of culture you like that that type of change so possible that sounds great uh so I read your Linkin profile in in in in all the details and there's a very interesting theme coming through that you know a lot of um um feedback from your colleagues are saying that you know you have absolutely no ego when it comes to do things like they're basically saying you know whatever it needs to be done you just go about doing this and that's you know people say that you're great great colleague and friend to them so I wonder and and and kind of uh simultaneously you Rose through the organization so I wonder and so it shows that really caring about what you do uh not your ego really you know is beneficial for everybody so but it's kind of I think it's it's hard for many people to let go it's a it's you know it's a highly charged environment uh so the question is uh can you offer some advice what like uh you know can somebody learn basically to let go and focus on the task and kind of forgo politics what what does it take to to do that first of all I'm flattered um I I I uh I think I think I think it's it's it's as as you pointed out it's it's non-trivial especially when you know you we live in these Matrix organizations with different teams and different you know potentially product areas where you know we're on a day-to-day basis focused on the thing that we like on our strategy the set of things that we have to do today or the next few weeks to to to win where winning is defined as you know some some objective but but it's very much local winning right like my boss or my boss's boss or my or me myself if I'm the leader like or us as a team Collective we decided this is what's going to be the best thing for our product area turns out all your your peer organizations again whether whether it's because you're Matrix or because you just have different teams they have their own objectives as well and they're trying to win and and many times those objectives are in Conflict right uh and I think the way that we resolve these conflicts ultimately is if we start from a a place where we where winning means what's best for the organization right it's not it's not necessarily what's best for me and that's and like that's better than what's best for you because you're in some other team and I know I have this ego or this passion I know this is right it's actually the reality is that these conflicting incentives and and and win strategies for local winning are are are very frequently in conflict and we have to work together to resolve those conflicts such that we can reach like a global Optima I think I think one other PE Point here is that uh sometimes as Leaders or as Engineers individuals like we're actually incapable of resolving these conflicts right like I have my you know I have my mandate you have your mandate we actually don't have the context or really the decision making authority to resolve the con conflict independently mhm like we we may be able to but many times we can't and and the right way to handle these situations again inside of uh what is optimizing for the global is is to to escalate it cleanly together to a person who has the the context and the authority to make the decision about how to balance these conflicting priorities if we if we treat if we think about it if if we behave in this way if we resolve conflict in this way if we always strive to think about you know the what's best for the company instead of what's best for me or what's best for the team I think that goes a really really long way that's I think that's a great uh great advice uh uh as as you know companies grow and Nitro is a growing company so I think the last question which I have is where dat driven company and you actually uh work with data driven products and Linkedin datadriven organization so we you know we're hiring data scientists we're hiring Engineers how do you uh make these teams work together how do you create this dat different products you know iterating fast and then startup fashion what's your take on this um it's a really good question I think it's I think it's hard I mean even even te companies uh at the size of LinkedIn and and teams at a company that's really data driven and I think has a data driven culture at LinkedIn struggle with this a little bit because especially in less mature products there's you have to balance you know gut with what the data is telling you and many times you don't even have the data you don't have sufficient data to inform your strategy like when we're talking about really mature product like profile at LinkedIn or growth at LinkedIn you have years in years of data and massive amounts of data to inform strategy in more venturous products like like pulse for example like we just don't have that much data and so a lot of it is is leveraging the data that we have and insights that we can extract from from uh from our users and and from their behavior and blending it with our Instinct about what we think our members want and user feedback like actually talking to the users and thinking about what value we're actually trying to create for them um and blending those two things together to create our you know both shortterm and long-term strategy MH um as far as how you build a datadriven culture um I think you know this this sounds a little bit circular but most people working in in techn technology and R&D organizations ultimately are very logical people and you know the basis of every logical argument is data so like even the most gut driven strategic Visionary product manager presented with you know uh with insights extracted from real data about how they're how the users of their product are behaving will be compelled in my opinion to leverage that and so you know if we push to to use data to inform our product decisions I think o over time that just changes the Rhythm and the way that we do things and how we think about developing product looking at the the opportunity that is available with one feature and the cost for example of implementing that feature comparing it with others to inform our priorization Etc okay and uh uh in terms of making data scientists and Engineers working together what's your Insight from LinkedIn experience uh they they have to work super closely like um the teams that I've seen flourish at LinkedIn and perform best are the ones where data science engineering and product and honestly design are all working very very closely together at every step of the R&D process from ideation as we talk about which features we want to prioritize and what the cost is and all that stuff to you know making sure we have the right instrumentation in place uh before the feature is shipped to then doing the followon analysis to determine you know is the feature or the product we ship living up to the expectations that we set for it at the ideation stage and then learning from that data to take the next step and iterate on top of it and if if all of these disciplines aren't working together then I actually believe you what you wind up producing are products that fall short of uh their potential thank you very much Shon that's great advice and we're looking forward to your talk cool awesome thank you Alexi