Episode 228: How People Analytics and Science Powers Amazon's Global Hiring Engine (Interview with Ashish Parulekar)
What does it take to scale talent acquisition at one of the world’s largest companies? When hiring spans hundreds of thousands of roles - from warehouses to tech hubs - the answer isn’t just more recruiters. It’s smarter systems, better data, and deep analytics.
In this episode, David Green is joined by Ashish Parulekar, Director of Data Science and Global Head of Talent Acquisition Analytics at Amazon, to explore how one of the most complex hiring engines in the world runs on data-led decisions.
Join them as they uncover:
How Amazon balances cost, speed, quality, and fairness in hiring
The tools and tech optimising high-volume and corporate recruitment
The various ways to assess top talent without bias
The evolving role of people analytics in unlocking productivity
What skills matter most in this growing field of people analytics
And how to avoid the trap of “data theatre” in early-stage analytics functions
Whether you're scaling a global team or building a people analytics capability from the ground up, this episode, sponsored by Worklytics, is packed with practical insight and strategic takeaways.
Worklytics helps leaders understand how work actually happens with data-driven insights into collaboration, productivity, and AI adoption.
By analysing real work patterns - from meetings to tool usage - they empower teams to work smarter, not harder.
And here’s something special: Worklytics is offering Digital HR Leaders listeners a complimentary AI adoption assessment to understand how your teams are really using AI - and where untapped potential lies. But don’t wait - spots are limited.
Learn more at worklytics.co/ai
[0:00:00] David Green: What does it take to hire at scale, when scale means hundreds of thousands of roles globally every year across fulfilment centres, corporate offices, tech hubs, headquarters, and warehouse facilities? And how do you do it efficiently, effectively, fairly, and powered by data from start to finish? I'm David Green, and today on the Digital HR Leaders Podcast, I'm joined by Ashish Parulekar, Director of Data Science and Global Head of Talent Acquisition Analytics at Amazon to explore exactly that. With a career steeped in analytics and a front row seat to one of the most complex and sophisticated hiring machines in the world, Ashish brings a unique perspective on how to make people data work at both volume and depth.
In our conversation, we dive into how Amazon approaches the twin challenges of high volume and high-stakes hiring, how analytics is used to optimise cost, quality, and fairness, and what it really takes to assess top talent without bias. We also look beyond Amazon to the broader people analytics profession. What skills matter most? Where do the biggest opportunities lie? And how do organisations just getting started avoid falling into the trap of data theatre, and instead focus on driving real impact? So, if you're leading talent strategy, building or scaling your analytics capability, or just wondering how to do more with less and do it smarter, this episode is one for you. So, without further ado let's get the conversation started.
Ashish, thanks for joining me today. To start the conversation off, could you share a little bit about your career journey that has led you to where you are today and your role at Amazon?
[0:01:54] Ashish Parulekar: Thank you, David, first of all, for having me on your podcast and all the great work you do for the field. I'm a big fan and really glad to be on your show. So, a little bit about me. So, when I did electrical engineering and data science, I did not anticipate being in the people analytics space. But early part of my career was exploration of different roles like data science, marketing, product. And then, it was really some key mentoring conversations that drove me to where I am, and I'll talk to you about the one about people analytics. So, I had been working in product for about ten years, really enjoying the talking to the customers, understanding their needs, building tangible things that help people and make a positive impact, and also bring in technology and analytics to do the good work in that space. And I had a few options for different roles. And I was having a conversation with my mentor on like, "Which one should I go with?" And she said, "All these roles are great. But, Ashish, when I see sparkle in your eyes is when you're talking about analytics, and when you're talking about people. And by the way, there is a role for that called people analytics". And I was like, "Wait, HR?" And she caught me by surprise.
It's funny how sometimes others know you better than you do. And she said, "You're trying to change your company's hiring strategies and performance management, and it's not even your day job. And you bring so much energy to those work streams that you should really look into it". And I respect her a lot, so I looked into it, and the more I learned, the more excited I got.
[0:03:37] David Green: Well, you've hit a very, very prescient advice that you got there. So, you were clearly a natural fit for people analytics. So, when you found out more about people analytics, what really attracted you to the field? And maybe an add-on question there is, what advantages do you feel that you had from the working outside HR, outside people analytics before, some of the skillsets that you could bring into the field?
[0:04:02] Ashish Parulekar: No, that's a great, great question. I think about this in three factors. One is purpose; then, the ability to make impact and the scale of impact you can have; and then, do you have the skill sets to actually make that impact, right? So, if you think about purpose, things like fairness go deep with people. Everybody know, everybody I think has a story of where they felt being wronged and things were not really meritorious and so on, and I have that story too. So, when you're working in a space that can help people get opportunities that are based on their merit and things that are more fair, I think there's a sense of purpose to it and sense of mission to it that you may not find in a lot of other spaces. And for me, that is the number one factor that got me interested in people analytics.
[0:04:56] David Green: We're going to talk about the work that you're doing at Amazon. Now, obviously, everyone listening to this show will know that Amazon is one of the largest organisations in the world. And I can only imagine what the scope and volume of hiring will be. And obviously, in your role as Global Head of Talent Acquisition Analytics, you're looking at all the data that supports that process. How does the scale and scope of talent acquisition at Amazon differ from your previous roles?
[0:05:26] Ashish Parulekar: So, scale is obviously, we are hiring hundreds of thousands of hourly employees, tens of thousands of corporate employees each year, right? And that might be the size of entire companies in many spaces. But I do think that the biggest difference is having skin in the game. What I mean by that is as my team is building models and putting them into production that drive marketing, that drive the funnel, that are hiring, so for example in hourly hiring, if our models are wrong, or if they are down because of a system issue, that directly impacts if we hire that exact number of people we need to hire at a location, which impacts if packages get delivered to people on time, which impacts the bottom line of Amazon. So, that direct accountability for something that you do that impacts the company's bottom line is very different. In many other places, you might have people that are interested in doing reporting or bringing interesting insights, then the operations team have to take those insights and make it a reality.
Here, teams are structured for single-thread ownership, and what you build, you put into production and it affects things in market, right? Even on the corporate side, we ran six million assessments last year, and this system has to be on globally 24/7. And if it's down for an hour, it can affect tens of thousands, lots of different businesses like Prime, AWS, devices, and you might lose out on a really important candidate who could build the next big thing. I think that accountability is a big difference. But with that accountability, also comes lots of benefits that make it a pleasure to work here, right?
With scale, you can actually do advanced analytics and science, and show what matters versus what doesn't matter, what are the myths versus what are the realities. With the underlying tech stack that AWS offers us, you can have sensors everywhere and make sure no data is lost in the ether, right? You have reliable, always-available data, which is massive if you're trying to do science. And also, the company culture being that of data-backed decision-making is really critical. That was one of the main questions I had when I was switching from the business side over to people analytics is like, okay, even if you come up with great insights, are people, are executives, are companies really going to make a change? Or are they going to say like, "Yeah, yeah, in theory, data-backed decision-making, all of us should do it, but in my unique situation, it doesn't apply. My judgment is more important and there is tremendous room for judgment always with people". But there are things where numbers can get you more right than wrong, and we should leverage both.
[0:08:30] David Green: What are some of the innovations or tools that you've implemented to optimise Amazon's high-volume hiring processes?
[0:09:36] Ashish Parulekar: I'm happy to talk about that. Things like AWS Connect help us build our own ATS system that can scale and have the fungibility and quick iteration that we need, as well as things like SageMaker, help us build models and put them in production. But let me digress a little bit, because it's most important to solve the right problem, right, and then the technology should be at the end of the journey. So, let me talk about, "Hey, what is the problem?" So, we have 3,000-plus sites around the world. And every week, a site might tell me, "Hey, I need five hires. The week after that, I might need 233 hires". And the week after that, that site may need 37 hires. And the site next to it might say 150 hires. No, 45 hires and 555 hires, right? So, you have 3,000 distinct strings of demand coming through. And then, we have three weeks of heads-up to know how much demand will be. So, you have a short heads-up time, we have unique demand, and then you have the scale of Amazon, right? So, that's the problem on one side. On the other side, I have 16 different dials or levers I can pull to get exactly the number of hires, and I need to get 95% accuracy in this operation. So now, all these moving pieces and your head can start spinning. So, that's the problem.
Now let's think about the context, right? So, six years ago, Amazon used to outsource this hiring, and we didn't have an internal team hiring in the hourly space. Six years ago, we decided, "Okay, let's bring it in house". And in the first year, we needed to double the hiring. And imagine doing it for the first time and you have to double. And then in the next two years, we had to go 6X, right? Massive growth. And a lot of that had to be built based on judgment and subject matter experts. So, about three years ago when I joined, we had this massive operation. For example, in marketing, we had 300 experts sitting in different media markets and telling us when to start, when to stop marketing, what channels, how much to spend. And we didn't have all this data captured. So, we know we are getting the hires we need, but we don't know the intelligence behind, like, what is the theme? How are we really making decisions? Where are we making good decisions versus not?
So, the first step was actually capturing all the data. A lot of the times people really underestimate investment in data and get distracted by shiny AI topics. If you don't invest in data, you are really capping the upside of what you can achieve. So, the first year was really about capturing data on what are the inputs people are using to make decisions, how are they making decisions, when do the outcomes actually pan out versus not. And the next step was just to put some simple rules around it. Like, what is the common theme across our whole network when certain decisions work versus others don't? So, some of the decisions started getting automated through rules, there was still a lot of judgment involved, right? And that alone saved or reduce our cost by 50%. And then, the year later, we were able to reduce the cost by 90%, by taking the rules and replacing them with machine learning.
So, there is this journey of capture the data, simple things first, put it into market, even if it's not perfect, learn from it, and then go to more advanced machine learning models. And even with machine learning models, it's not fully automated, you still have judgment in the right spots. So, we've done that transition over time. And when you come to that final stage, and that's where I can talk about our ML Ops and SageMaker and how it can help us run these models globally for these 3000 sites and the thousands of hires that we are making that scale, and be accurate. But it comes at the end of the journey of prototype and integrate over time.
[0:14:11] David Green: Let's just switch a little bit to corporate hiring. Now, you obviously do a lot on the corporate hiring side at Amazon, but obviously not anywhere near the volume of what you're doing on the hourly-paid workers. What are you currently doing to ensure that you are acquiring the top talent in the space, but also assessing them in a fair and unbiased manner?
[0:14:32] Ashish Parulekar: Let me start personally by saying, my mission is to minimise guesswork and bias when it comes to people decisions. That's what drives me, right? And corporate hiring space is where that is probably most applicable. So, if you think about our hiring experience today, for most companies, from candidate perspective, they feel that they are tossing their resume on a pile of thousands of resumes, they don't know if anybody's even going to take a look at it. And probably, their baseline expectation is not to hear back from anyone when you apply online, right? And they're told, "Okay, you need to network with the hiring manager or the recruiter, etc". And then, there's a lot of luck involved in if you even get a fair shot at evaluating what you can do, and how does that relate to a job.
Then, on the flipside, if you're thinking about the recruiter, you've got stacks of thousands of resumes, right? It's not practical for you to really evaluate all thousand resumes, right? So, you put some filters and narrow things down, and you're able to maybe look at a few hundred, 15 to 30 seconds at a time, right? And in that time, it's a really tough job to pick out of that pile really the top candidates. And that's where, again, human judgment and machines together can do a better job. So, we have automated resume reviews, we have online assessments that we have scaled for hundreds of job families. So, we have 455 job families for which we have online assessments. We had six million-plus assessments last year. And the thing I love about assessments is, it's an opportunity for a candidate to show what they can do. Here are the skills that are important for the job; show me what you can do. And so, we gave people six million-plus merit-based shots to get a job at Amazon. And for some candidates, that can truly be life changing. And it could be life changing for the company and our customers, because if you hire the top talent who is best suited for that job, then we can make a bigger difference for the customers.
We also have machines helping candidates navigate, so each company has different titles and a different way they've organised functions. So, when candidates come in, they can upload a resume, and we can help them direct like, "Hey, based on your skills, here are actually the jobs you might be best suited for, here are the levels and families". So, there's a lot of technology that can help improve both the candidate experience as well as the recruiter experience, to create the best match in the end between the job and the person. And then, that's really what we are going for in the corporate space.
[0:17:33] David Green: Very good. Six million assessments, wow!
[0:17:36] Ashish Parulekar: Yeah.
[0:17:37] David Green: That's a lot of data as well, obviously.
[0:17:39] Ashish Parulekar: Yes, that's a ton of data, but it also requires some incredible subject matter expertise in things like IO psychology, as well as engineering, to make sure you can deploy these assessments and keep it up 24/7 all through the year. So, a lot of amazing work that the team is doing.
[0:18:00] David Green: Yeah, really good. So, on the topic of corporate hiring, let's turn the lens on people analytics itself, which in some respects is still an emerging field, still growing. What would you say to someone considering a career in this area? And then, by extension, what would you say the top skills that you look for when hiring people analytics professionals?
[0:18:24] Ashish Parulekar: Yeah. So, my pitch for anybody who's maybe on the tech side, business side, to come over to people analytics is the same as we talked at the top. So, "Hey, this is a really mission-driven field. You can make 10x, 100x difference in this field. And a lot of the skills that you learn on the business side and the tech side can translate very nicely in this field. So, it can be a really fulfilling endeavour, and you can be in the frontier of lots of problems that haven't been solved yet". So, that would be my pitch for somebody considering. And especially at Amazon, you have the scale of data to actually do the similar types of analytics and science that you could do on the business side.
In terms of what I would advise people in general, if they're thinking about career in this space, and may not be too different from career in other spaces, is I feel, going forward, big wins are going to be at intersection of functions. So, let's say data engineering, right? I feel especially, not just in people analytics, but with the advances in AI, a key skill that everybody's going to need is data engineering, right? And it's quite scarce, actually. We find that finding great data engineers is harder than finding great scientists, just based on the number of professionals in the space. But if you are a great data engineer with advances in tooling from places like AWS and what AI can do these days, going deep in data engineering alone may not be enough. And we are going to need some people who are incredibly deep in data engineering. But for most people, if you can marry your skills of data engineering by also understanding the process, understanding the business, then you can make the right decisions on where to put the sensors, what is it that you need to capture, how do you prioritise work, right? And that's going to be most valuable to your end customer, that's what's going to be helping you make the most impact. So, intersection of those two things.
Let's talk about science, right? I talked about like in selection space, you need deep expertise in IO psychology. But if you marry that with ML, you can scale so much faster, right? Like, one of the biggest challenges I've heard in the time I've been in the people analytics space is like, "It takes too long for us to build selection mechanisms, like, it takes 18 months, right? Oh my God, the world changes in 18 months, right?" But how can we marry multiple, so we had automated resume review for 20 job families last year, and now we have it for 220, right? Similar scale of growth in other areas, like online assessments. So, how can we marry different sciences together to scale? In product, I think that's where probably three intersections come in. I think product as a discipline is really hard to find great product managers. But if you can do product and software and science, that's a killer combination. And if you think about where the world is going, you are going to need all these three skills to really leverage, harness the goodness that AI can bring you. It's going to be science, it's going to be software, and it's going to be product.
So, in general, I would say intersection of things is where big wins are. So, if you're early in your career, explore versus maximize. If you just stay in one function and maximize, you may be limiting the upside long term. And you might be better off early in your career going broader and collecting a broader set of skills, so that you are building a good foundation for a greater future in the long term.
[0:22:31] David Green: Interestingly, maybe it's just two parts of this question as well. What roles are you currently hiring for on your team at Amazon? And then, the second part of that, maybe how the skills and capabilities are evolving over time, but the second part is also, where do you envisage, maybe in the next 12 months, 18 months that you might be hiring additional skills, maybe new skills into your team?
[0:24:08] Ashish Parulekar: Yeah, so we have 25 roles open right now. Given the size of the team, we typically hire about 50 people each year. We've got roles in data engineering, as I said, we have a large presence in data engineering globally; we've got software engineers for deploying things like assessments and automated resume reviews; we also have IO psychology roles, as well as data science and machine learning roles. As always, I always would look for product people who can do all three, like product, science and software. There is always a need for that. So, you would have a range of roles available. And I would invite both people who are in people analytics for a long time, as well as who are on the business side, to come take a look, because one of the hesitations people may have from moving on from the business side, is the scale of data and access to technology that you may have on the business side, do you have that on HR side as well? And at a company like Amazon, you do. So, if you're considering a change, this might be a good place to start, and then you can go from there.
[0:25:31] David Green: And if we can talk a little bit about the product management piece, because again, it's an area that we've noticed growing in many people analytics functions around the world. We've identified a gap in the research that we've done around democratising data and the adoption of those products as well. And product management plays a really important role in that, doesn't it, not just about user experience but it's around making the products. Talk a little bit to the product management piece, that'd be great.
[0:26:03] Ashish Parulekar: So, on one side, if you think about the users, they are on this journey about data literacy, and then comes science literacy. So, what they deeply know is the problems they haven't had today, right? But if you ask them for solutions, they might just tell you, "Hey, help me do what I do faster or easier". But you may not get the transformative idea from your users. So, as a product professional, you have to understand the need, but then also understand the technology to see what is possible to come up with a function, come up with a solution. And then, you have to sell that solution to the customer. That's also really challenging because they might not believe you, right? And in the end, as I said, they are on the hook to deliver. So, you paint a picture which may look like a crystal ball, and they might not sign up for that solution. So, how do you bridge? You deeply need to understand the technology so that you're not promising a crystal ball, you're promising something real; and at the same time, you have to sell it to the customer so that they believe it.
Then, once you have that alignment, you have to go build it. And in building it, there is such a dramatic curve of improvement in technology. What used to be technical debt over years, you can accumulate technical debt within months now, right? The things you're building your infrastructure on can be obsolete within months. So, how do you also stay on the frontier of technology, where things are going, to make the right investment decision and build things right, give yourself optionality to be on the most modern technology? That's the other side of the story. So, really incredible opportunity for people who can do that. They can really unlock the potential of all the individual functional experts we have if we have great product innovation.
[0:28:24] David Green: So, as someone who's been deeply involved and embedded in talent acquisition analytics, and I know you were doing similar before you came to Amazon as well, what do you see as the biggest opportunities for people analytics to drive even further impact with regards to talent acquisition?
[0:28:42] Ashish Parulekar: I think in the talent acquisition space, the biggest problem is quality of hire, right? If we can nail that, I think businesses would be willing to invest a lot more in this space. And it's about measuring the quality of hire and then showing you can actually move the needle. In general, even if you are just doing, let's say, skill-based hiring and using best practices, you're probably in the top half of the companies anyway. There are a lot of companies who are still not doing skill-based hiring, they're not doing structured interviews. These are proven practices for decades now. So, if you're not doing that, do that, right? I think that would be the first step.
But even if it's skill-based hiring, and given the limited time we have to evaluate talent, a few hours, getting skill-based signals is probably a good way right now. But that still leaves the unknown of, can this candidate put these skills together to deliver the tasks that they need to complete in the environment of this company, in the culture of this company. So, that remains unknown. And a lot of the misses we end up having are because people may have the underlying skills, but cannot put it together, or maybe cannot put it together in this environment. But with AI, I believe we can make real-life job previews happen. And in that environment, it's a great experience for the candidate to see how it's really like to work in a particular role in a particular company, what they would be asked to do day-to-day, and can they pull all the skills they have together, both the soft skills and the functional skills, to deliver the solutions that the company needs? And that's where I believe the world is going.
If you're able to do that, then you will be able to measure more accurately a person's ability to do high-quality work in that particular job. And it would be better for the candidates as well, because it's a life-changing decision to change a job, move potentially to another place. And then, you have all these unknowns that you don't know how are going to pan out. So, the closer we can get to a real-life job preview, and I believe with AI we can, I think that would be the next frontier.
[0:31:24] David Green: If we look more generally at the field of people analytics, and this leads quite nicely, actually one of the key topics that business leaders talk about is productivity. We've discussed it in the past. You posted some excellent articles on it a few years ago as well. But for the benefit of listeners, what role can people analytics play, in your opinion, in terms of unlocking productivity at the individual, the team, and maybe the organisational level as well?
[0:31:52] Ashish Parulekar: Great question. And these days, it's pretty easy to jump to AI as the answer for productivity. And I always say, it's most important for us to solve the right problem first and get the structure right, and the technology comes at the end. Like, don't start with technology, right? And speaking more broadly of people analytics, a big part of people analytics is consulting with the business. So, if you're thinking about organisational level productivity, I still think it is more about, how do we make sure we are solving the right problems? How do we make sure we have the right people to solve those problems? How do we make sure those people have the resources to solve the problem? And so on and so forth. So, making sure we have that structure right. And a lot of those things might be more about having clear decision-making mechanisms, having a great way to do workforce planning to make sure your top talent with the right skills are in the right roles. And there are areas where AI plays a role in that. But it's not just about AI.
When I talked about, we cut costs in hourly hiring by 50% and improved accuracy in the high 90s, that had nothing to do with AI, right? So, it's really important not to be jumping to that end state. Do the right problem-solving techniques, like break down the big, complex productivity problem into components, make sure that you can solve each of the components, make sure you understand the ecosystem and you can do system-thinking to figure out what you should solve first before you solve next, and what is the knock-on effect of solving one thing on another. And then, be aware of what AI can do in those steps. So, be aware of AI and apply that tool when it's the right tool for the job, versus using AI, hammer looking for a nail. That's probably not the right way to approach productivity.
[0:34:15] David Green: It's really interesting, Ashish, because I mean when Jonathan and I were writing Excellence in People Analytics and prior to that, and we weren't the only people saying it as well, I'm not going to claim ownership on this, when people analytics was maybe in the mid-2010s, lots of people wanted to jump to the fanciest type of analytics like," I want to do organisational network analytics". We'd always say, "Well, what's the problem you're trying to solve?" As you said, properly define what the problem is, maybe have some questions that you want to answer, some hypotheses that you want to test. And only then, once you've gathered the data together, do you start thinking about, "Oh, what's the solution for this?" And that's when AI comes in there, and not, as you said, not a hammer looking for a nail. I like that analogy.
[0:35:02] Ashish Parulekar: I'll give you an example. In one of the early roles at Capital One, I inherited a team that was doing people analytics for our software organisation. And the team was quite burnt out, and the customers were not very happy about the outcomes. And the team had received 400 different requests for projects through the year, the previous year before I joined. And they had all great science and data engineering, like they had applied the best available tools at their disposal. But nobody was happy, right? So, I was like, okay, in that situation, try to go as high up as possible in terms of your customer. So, I went to the CIO and asked him, "Okay, you're funding this team. Tell me, because of the work we have done, is there something you're doing differently this year compared to last year?" and he could claim nothing. There was nothing that the tech organisation was doing differently because of the 400 things we did last year. And the team is burned out because nobody's happy.
Now, I said, "Okay, all right, so what if we did two things, and really made a difference in two things, what would be your two things?" At that time, it was about hiring and retention. And I went, "All right, so we're going to staff 60% of the team on these two problems". And then, there's always this executive concierge, questions from up top that come, and you've got to have like 10% for that, and then the rest is KPRO. So, if that is what we did, we might get two, that's better than zero. Now, this is a risky endeavour, right? You might have this conversation, and you might be like, "Okay, I actually don't need this team, because I haven't done anything different". So, it's a delicate conversation. But then, at the end of it, when we actually focused on those two problems, now we were able to increase the throughput of software engineering by 50%. But just focusing on that piece, solving that problem deeply.
Then, there were conversations about, "Okay, how can we fund this team for a third priority? So, really, if we had stayed in this, "Solve 400 problems with top technology", we may not have actually made any impact. So, that focus, solving the problem, the right problem, the right way, and maybe AI is the answer to some of those problems, maybe it's not. Sometimes, four-week average can solve your problem, and you don't need AI.
[0:37:36] David Green: It's really important, Ashish. I mean, so many times we hear about analytics teams that are burning out because they're just doing too much, but they're not having much impact. And it's because they're doing too much and not prioritising the right things. And that's such a great example of doing that. And that's, again, how you build trust with your internal customers, isn't it? What are their key challenges they're trying to solve for? Help them solve for them, they'll work with you more, and then actually it's the business that will help you get more investment in the team, whether it's people, technology, other resources. So, yeah, really good example there as well.
So, I'm going to go to the question of the series, Ashish, now, and then I'm going to come back and just ask you maybe to give some key things for people to take away for them. So, the question of the series, for those first-time listeners, this is something we ask everyone in a series of the Digital HR Leaders podcast, and it really is, we're going to talk about AI a little bit here, Ashish. How can HR use AI to improve employee experience and wellbeing? And if you want to extend that into candidate experience, please feel free to do so.
[0:38:47] Ashish Parulekar: Yeah. The one way we are thinking about this is, where is human touch truly needed and truly beneficial, and what is taking away from humans being able to spend that time in that human touch? So, think about recruiting. On one side, there's a candidate who could be making a life-changing decision, could be uprooting their family and moving somewhere, or being at a job they're doing really well, but they're interested in something different. These are big decisions. And it's really important to have somebody you can trust on the other side. And that is really where humans can differentiate. At least personally for me, talking to a machine may not feel the same way. But recruiters today have do so much admin work, right? Like 30%, 40% of their plate can be admin work, and that takes them away from being connected to the candidate. And how can AI take off that stuff from that plate to make sure, not just in, let's say, operations roles, even in data engineering, right?
I remember back in the day, working on mainframe and other technologies, it was a bear to understand where your data is, where it's coming from. Today, with AWS, it's like having a conversation with your data, to understand the metadata behind, where is this field, what does it mean, what is the range, what is the source, what is the latency? I can have a conversation with my data. That is just phenomenal experience for a data engineer. And now, they are freed up to do the engineering work they need to do versus the admin work of just understanding the basics.
[0:40:38] David Green: We've got to the last question, Ashish. I really enjoyed this conversation and I think hopefully, listeners will really be able to take a lot from it. But for those people listening that are working in organisations that are maybe just beginning their journey to embed analytics into recruitment or talent acquisition, or maybe want to take it to the next level, what advice would you give? Where should they start and what are the key success factors? And I appreciate that the same success factors don't necessarily apply to every organisation.
[0:41:09] Ashish Parulekar: So, I think there are a few different patterns, right? So, one pretty well-known pattern is types of analytics. So, you could have descriptive, like what is happening, right? Diagnostic, like why is it happening? Predictive, like what will happen? And then prescriptive, what should I do about it? So, people analytics teams evolve in that direction as they have more time under their belt, more data, more size of the team, they can move. Usually, everybody starts with descriptive, right? Now, that is one model. Now, the thing I would add to that is, rather than going descriptive on everything, a lot of people analytics teams I see do tons of reporting for everything under the sun, so they are taking the descriptive piece and going really broad. I would actually highly recommend people to go deep in few problems. So, don't do reporting for everything. Pick two problems or one problem that business really cares about solving this year, and go deeper in that problem and make a difference in that problem. Because let's say this year, it might be about workforce planning. If you can really make a difference there, then your team will get funded to solve another problem, and another problem. But from the business perspective, it's really about, are you making an impact at the bottom line?
[0:42:42] David Green: Before we wrap up, where can listeners find out more about you and learn about the work you and your team are doing at Amazon? And maybe, is there somewhere they can go if someone's thinking, "Okay, I'd like to work with Ashish and his team", as well?
[0:42:56] Ashish Parulekar: Yeah, so best way to find me is on LinkedIn. And best way to work with us is go to the Amazon website and search ITA, Intelligent Talent Acquisition. So, if you search ITA in quotes, you'll get the direct match for 25 roles that we are hiring for today. I'm sure there'll be more coming up very soon.
[0:43:18] David Green: Brilliant, and we'll put that in the show notes. But basically, go to the Amazon site, search ITA, Intelligent Talent Acquisition, and you'll see the roles that Ashish was talking about earlier. So, Ashish, thank you very much. I look forward to bumping into you hopefully at a conference in the next few months as well.
[0:43:37] Ashish Parulekar: And thank you so much for having me.