Joe Fuller: Despite widespread and often widely publicized efforts to diversify the technology workforce, the sector remains unrepresentative of the wider population. And the pandemic seems to have increased disparities in opportunity and income. What’s behind the failure of the system to change? And how can employers, educators, and nonprofits make progress against this important goal?

Welcome to the Managing the Future of Work podcast from Harvard Business School. I’m your host, Harvard Business School Professor and Visiting Fellow at the American Enterprise Institute, Joe Fuller. I’m joined today by Michael Ellison, Co-Founder and CEO of CodePath. The nonprofit provides skills and job opportunities for computer science students from low-income and underrepresented minority backgrounds. CodePath works with colleges and employers to address the skills gap and introduce a higher level of diversity in IT and related fields. Through curriculum support, training, and paid internships, CodePath readies participants for high-skilled jobs in IT, such as software engineer. We’ll talk about CodePath’s success in the classroom and the workplace. We’ll also talk about what it takes to provide the combination of hard and so-called “soft skills” that can secure its graduates highly paid jobs that are great springboards to further career advancement. Welcome to the podcast, Michael.

Michael Ellison: Thanks. Thanks for having me.

Fuller: Michael, you’ve had a diverse career. You’ve been in start-ups, you’ve been on the technology side of venture capital, and now you’ve started CodePath. Tell us a little bit about how CodePath came into being.

Ellison: Yeah. Well, a lot of it just starts with my personal story. I grew up low-income, rural Maine, single-mother household. We were homeless at times as I was growing up. It started when I was five years old, my father was incarcerated, which then was the catalyst for a whole lot of different challenges, moving around quite a bit. Sometimes you’re in better schools, sometimes you’re in schools that aren’t quite so good. I think I was really fortunate to be in Maine, because, in general, there’s a lot of really good schools, a lot of good teachers. You can be low-income, but you can still do pretty well. When I got into college, it’s where I caught the entrepreneurship bug, where I started a first nonprofit when I was 19 years old. I founded several additional nonprofits, trying to figure out how you could scale social impact work, especially in education.

Fuller: Michael, you had technology jobs early in your career. You were part of a founding team of a company that enjoyed a successful exit. What did you learn during that time, particularly in your early career? And how did that inform the way you think about serving your target population at CodePath?

Ellison: Getting to tech kind of taught me about how you build large organizations, you understand incentives and systems to drive change, and was fortunate to be on a founding team that was acquired for a little over $3 billion in 2020, which even further cemented to me how critical it is that we are trying to leverage our education system to get young people from similar backgrounds into competitive, life-changing roles and positions and opportunities. So all of that informs CodePath’s approach, where we’re an education nonprofit. We’re focusing on increasing diversity in the tech sector, because we see that as the pathway to upward mobility. It’s the number one source of new wages—50 percent of the highest-paid jobs are computing jobs.

Fuller: CodePath works primarily with people in postsecondary education—community college students and particularly students at four-year colleges and universities. How did your own college experience influence the design of CodePath its mission, how you thought about it?

Ellison: CodePath has become a layer on top of colleges and universities. We’re changing what’s being taught, how it’s being taught, training professors, changing curriculum. And you can think of the intervention across each of the different places, the root causes, where often people from low-income, underrepresented backgrounds will fall off track, while CodePath keeps them in. So that means freshman year, courses that you would take before your first introductory technology courses to give you confidence to build the skills that make you better prepared to persist going to sophomore, junior year. Also addressing the critical knowledge and skill gaps to ensure that when you get that first technical internship or that first job, then you’re prepared to have outsized achievement, as well as also having that confidence, as opposed to oftentimes young people are coming unprepared even though they’re doing well academically. So all of that holistically is the focus.

Fuller: So, Michael, we have made a study in our project, Managing the Future of Work, of different interventions of different types for skills building to improve DEI, and there’s no shortage of social philanthropists who are trying to create platforms for diverse applicants to get into the stream of technology jobs. What is it that CodePath’s bringing that’s different? What did you see when you looked at the landscape and said, “I think there’s a different way or a way that would really augment what’s already being done effectively?”

Ellison: We saw two big, major gaps—one being a lot of programs focusing on inspiration versus the entire pathway, the holistic, filling all the different skills. You see, for example, a lot of different boot camp programs, where they’re very, very focused on one framework, one technology, without necessarily having the wraparound, so that a lot of the people applying to those programs and getting in already have college degrees or already have technical degrees. Number two is our emphasis on very competitive, very high-paying roles, where many of the organizations in our space, they focus on help desk IT entry-level. But you’re talking about huge differences from salaries. And you look at help desk IT, $50,000, $60,000, which is great. But then a difference with software engineering, closer to $90,000 on average, or you look at the most competitive software engineering roles, $150,000, $200,000-plus entry-level starting salaries.

Fuller: So presumably you’ve reviewed this through the lens of what you think the barriers are that confront people of color, diverse communities, from getting on a pathway to those higher-echelon tech jobs. What do you see those barriers as being, and how does your design seek to circumvent them?

Ellison: It’s not just the underrepresented minorities. We’re also looking at low-income, first-generation college students. That intersectionality is really, really important. I mentioned inspiration as being key, but then the gap between, “I don’t know anything,” “I don’t feel like tech is for me,” versus “I’m in one of these most competitive roles,” it’s not a couple of hundred hours, it’s thousands of hours. So we just don’t pretend like we can, in a couple weeks, get someone to that level of proficiency. We’re also augmenting and changing hundreds of hours of what that curriculum and student experience looks like.

Fuller: You used an interesting phrase, that initially CodePath kind of rides on top of colleges. Say more about that. Where were they falling short? Or is it their understanding of the state of the art? Is it their pedagogy? Is it the completeness of their programs? What are you bringing to add to what they can provide to get people on these pathways?

Ellison: So across higher ed in computer science, there’s a shortage of CS professors, and then there’s also certain incentives that a professor has in order to have career mobility—research and publications and so forth. Combine both of those, and then you end up having CS curriculum that may not change all that much in 10 or 20 years versus industry is changing every couple of months in terms of what they expect. You have a lack of support. So students from more marginalized communities need more support. They didn’t have the support in high school, they didn’t necessarily have exposure to Advanced Placement Computer Science [AP CS]. With CodePath, we’ve actually made it so that, by professors or schools working with us, it incentivizes the professors. We save a professor around 60 hours per semester if they work with us. We’re doing grading, we’re doing curriculum, student support. We don’t charge schools or professors. We also are helping to solve pain points in terms of their capacity. And from a quality standpoint, they see that very quickly the students participating in our program are getting jobs at companies that previously was hard for them to even get them to come to the campus. So we very much are thinking about those incentives to change the system, make it very turnkey, very low friction. And then that’s allowed us to scale to over 70 colleges and universities in just a couple of years.

Fuller: Well, let’s unpack that a little bit. When you’re approaching a school for the first time, how do you describe what you’re going to do and how do you approach the faculty, because I can tell you, I’m a professor of late vocation, but faculty are often pretty sensitive to the suggestion that what they’re doing isn’t working out or that they could use some help. Now, of course, if you’re saying I’ll give you 60 hours back a semester, you’ve got my attention. So we’ll talk offline about CodePath providing help for business school professors who teach general management. But tell me more about, what’s the pitch to one of these schools? And how have you refined it over time?

Ellison: Well, we’re entrepreneurial, and sometimes if you can’t go through the front door, you go through the back door—we’re thinking of creative ideas. This might surprise you: The first successful school that we were at—before CodePath was a nonprofit—was Stanford, and I was the one who was leading this. I called professors, I showed up, they didn’t want to talk to me; they cursed me out. They’re like, “Why do I want to change my classroom? I love my students. I love what I’m doing. I love how this is.” The reason why they showed up on our radar was because we were at the time teaching these courses for senior engineers, 8- or 10-years-plus professional experience, and Stanford students kept trying to get into our courses. We were like, “Why do you want to take our course? You have some of the best courses anywhere.” And they said, “No, no, no, no. We keep hearing from alumni that there’s gaps and that we need to know what you’re teaching.” So the professors didn’t listen to us. We started to go through student-led courses. That’s where our strategy to go bottom up, have hundreds of students as champions, before oftentimes professors hear about us. And then also the infrastructure we built around allowing it to be possible for students to have these really managing more of what looked like in-person lab sessions with a flipped classroom on top of that, and logging into our software platform, having professional engineers embedded in these different learning environments as well, to have that expertise that doesn’t necessarily need to just rely on one professor in that classroom. You don’t start from, “How do I convince someone who it’s really, really hard to convince?” but rather, “How do I already cross that trust gap through the students as your step one on a college campus?”

Fuller: So you pivoted pretty quickly, like a lot of entrepreneurs do, from your original hypothesis. How did you get beyond the Harvey Mudds and the Stanfords and the Northeasterns to schools that don’t have long histories of presence in computer science and deep faculties and access to large employers and whatnot?

Ellison: So, actually, it was an immediate pivot. We were only at Stanford one semester. We’re equity focused. We noticed that a lot of tech companies, where they talk about alternative pathways, but you often will see these diversity programs filled up with very wealthy students from a variety of backgrounds, but very wealthy. That’s the norm. And we know this from running internship and diversity programs from different tech companies. So we decided we would go away from Stanford and instead be very focused on the largest concentrations of our target population. So that’s how we do school selection—the largest numbers of Black, Latinx, Indigenous students in STEM, in CS, and then looking at opportunities where you can increase enrollment or decrease attrition or the variety of strategies that move the needle. And what we realized is that these schools go beyond the top 25, top 50 computer science (CS) schools. And then they’re so hungry for increased capacity. They’re struggling to hire enough professors to even do the bare minimum. And then you come along, you’re not charging them anything, you’re giving them training, you’re able to put them on the map with employers that they would love to have relationships with. And then the students are loving this experience, and they’re getting real job outcomes. And we’re talking about, some schools we’re working with, students are applying to hundreds of tech companies, they don’t hear back from anyone, and then suddenly they’re working with CodePath, and then students are getting offers from their Amazons and Googles and so forth. It’s a game changer, and what I mean when I talk about aligning incentives. And the schools that want to work with us the most are those ones that have the least resources.

Fuller: Let’s think about this as a three-stage process. There’s the decision to pursue a credential, a degree, in computer science; there’s seeing that through, completing the program; and then there’s actually getting placed. Have you seen patterns of barriers at each of those stages? And how do you design an intervention that helps people overcome those barriers when they do show up?

Ellison: There’s so much noise in our space. There’s so many start-ups and nonprofit and for-profit. Students are just trying to figure out how to do better, and they don’t know the difference between these different programs. They will respond to “This is free.” And CodePath is free. We do see lots of challenges where students are participating in different programs, and a lot of the MOOC—Massive Open Online Courses—connected with credentials, some have watered down their programming in order to make it easier to have higher completion rates. But then you have the learners that don’t actually know when they’re going to be able to get a job opportunity, or if this credential—is it one credential, is it 10 credentials? At what point are you ready to get that job? Do you need separate support on top of that? So we see a lot of those challenges there. But we also see community. Different CS departments say this is the way that you get a job in a top company. Other nonprofits in our space partner with us to be the difference between not being able to pass a technical interview and passing a technical interview. So we have a lot of these innovative methods. But I’d be lying if I said it wasn’t very challenging to compete for attention. You have, separately, the program, itself. Most curriculum and most training or educational programs—inside of companies, outside of companies—they are not personalized as well as they could to be that right size fit, but we care very much about that, and it correlates with really high completion rates. That last element to things, where the placement… Passing technical interviews is a separate skill set. And then we’ve also been for a couple of years now testing out strategies to, at scale, get a first interview or get a first job without requiring a resume. And when you talk about skills-based hiring, you have to be able to cross the trust gap. Schools like Harvard, Stanford, they’ve crossed that. Employers are willing to just, “Yep, come, I’ll talk to you. You probably have the combination I’m looking for.” We need to do that at scale for people who have the skills but don’t have the connections, who don’t have the pedigree.

Fuller: So one thing we hear in lots of efforts to drive better outcomes in the labor market—across skill sets and geographies and populations served—is this notion of surround services, that if you provide tutoring or some financial aid to a working learner or coaching in social skills, then that job interview or those critical first weeks on a job go better and success ratios jump up. It sounds like, in your own way, that’s integral to your approach. Do I have that right?

Ellison: I would say yes. There’s a variety of different services. We are analyzing different schools and how well they fill different gaps on the road to what we would call a proficient software engineer. Different schools do a better or worse job across different gaps, which means you might have slightly different programming depending upon the size of a CS program, the number of faculty, the wraparound support a campus has. But what I’m getting to is, what’s the proper dosage of these different services and interventions? It’s not all the same, depending on who you are. And that’s where the upfront assessment or understanding of where someone is and continuing to reevaluate where they are to recommend the right next step is so critical. It’s what leads to the right thing at the right time, where someone would’ve dropped out of tech, or they would’ve not had the skill to pass a technical interview. And maybe just a specific example: A lot of students will themselves study a lot of data structures and algorithms to get into a competitive tech company, when they don’t realize that they need to work most on technical communication and how they think out loud as they’re doing their problem solving, similar to a management consulting type of interview.

Fuller: It certainly addresses what’s been an abiding problem with a lot of interventions, which has inhibited scaling. If you assume every person you’re serving have to get the same offer, basically, then it becomes either very low common denominator or really expensive to provide. But if you can titrate the offer to scale it to augmenting the weaker parts of a curriculum or an offer or helping a student at those moments of truth where they’re going to be most at risk or where their natural aptitudes aren’t showing off to their best effect, that’s a very powerful model. Let’s jump to the demand side. All our research and our advocacy is very much about trying to get the voice of the employer into programs early and often, ideally with some commitment on their part to at least be supporting the framing of the intervention, so that there’s a lot of fit with what the demand side is seeking in applicants. And also we’re big advocates for compensated work-based learning: internships, apprenticeship, really substantive and paid. How are you interacting with employers? What have you been learning from them?

Ellison: It’s been very exciting over the past couple years. You have the pandemic, which has created a whole wealth of new different ways of working—from in-person apprenticeships to online remote apprenticeships, these various types of micro work opportunities. We’ve decided to create programs that can more effectively help companies to do, at scale, diversifying their workforce, hiring from these unconventional pathways. And one of the favorite initiatives that we have is, it’s a sophomore internship program and we observe that when you’re looking at big tech employers, 70 percent, 80 percent-plus of entry-level software engineers come through their internship program. Well, we’re a layer across colleges and universities. So what if we can create a program that combines the courses as a top of funnel to have us then control selection of the sophomore year internship that then can be part upskilling, part evaluation for the very competitive junior internship program that then results in, “Do I get a full-time offer or not?” So this was a hypothesis. It was based on, we’ve been running, Meta has a sophomore internship program, which is their primary diversity strategy. We redesigned that, have been running it since 2015, and in the past couple years, during the pandemic, we have created versions of a sophomore internship program that can be remote, it can be on site. We just created one, and we’re almost finished with the pilot with Salesforce.

Fuller: Well, certainly one thing—that magic word you said there, Michael, was “sophomore.” Sophomore internships, paid ones, are rare, and when you’ve got a sophomore, you also have someone who’s got two years of study left. So if they get a chance to be in your environment, they might learn things about what they should be putting their effort against in their junior and senior years. Or you might, of course, have them come back as a junior intern as well, if you’re hoping to win them back.

Ellison: And it could be anything. The idea here is that companies are willing to invest in multiple months of some training, upskilling, and I get to hire out of it. And right now, it’s a sophomore internship program. It could be for seniors; it could be for people who are not in the workforce. And when I look at the landscape, the workforce landscape, there’s a lot of these programs.

Fuller: I think our research would indicate that there’s been a big change in the last five years, which is that more companies are acknowledging that, in technology specifically, that their traditional approach to hiring talent is just not insufficient. It’s insufficient in the volume it generates, it’s insufficient in the quality they generate. Certainly some major companies I’ve talked to have said graduates from computer science programs in the United States show up, and they know a lot of theory, but they don’t know a lot of practicum, and that there’s an immediate additional education component to it, which is, how do you turn what you understand about networks and user interfaces or writing code or whatever else and turn it into skills that can be harnessed in a project setting in doing actual software engineering. So the companies are experimenting.

Ellison: When I chat with CIOs, they say things like, “I want athletes, full stack engineers—so multiple technologies across different areas of the stack. I also want them to know secure coding practices. I also want them to know multiple front-end technologies, DevOps, provisioning, secure. And then also on top of that, do they understand cloud? Do they have AWS certifications?”

Fuller: I think that what that does indicate is the amount of pressure that those executives are feeling, that they really want all-arounders who have that type of full array of skills around full stack engineering skills in important platforms. And what they have to understand is the fact that they’ve not had a sophisticated supply chain management—talent management supply chain—mentality for the last 10 years cannot not be made up for by saying, “Can you give me everything I wish I’d been growing along the way now,” because, of course, anyone with the skills you’re talking about is probably in their mid-’30s with 10 years of job experience. I think a metaphor we use is Minor League Baseball. We can give you prospects, and you have to grow your own in some regards and, yeah, there are some phenoms, but what are the things that are going to get that person in your setting, adding value so they can gain experiential learning, on the job learning, and learning that you are going to provide, or that others will provide, other skills providers: the 2Us, the Courseras, or someone else might provide down the road. I think that it is indicative of the tension also between engineering organizations, CIO organizations, and the HR organization.

Ellison: Yeah. And when we do partnerships with companies, it is very often a partnership with HR and with engineering. If the most senior engineering leader isn’t excited, then there’s a good chance you’re probably not giving them what they want, what the organization really wants. And then HR, of course, this is their domain. They need to have ownership, they need to feel like this is a value add. And then if you can get both of those right, then that’s our hypothesis how you get to some at-scale outcomes and interventions.

Fuller: How do degree-granting programs fit into the future? Have we seen the highest degree of penetration importance applied to four-year degrees, and is that going to recede now? Is a four-year degree going to become more and more obsolete? How do you see that?

Ellison: I don’t think it will become obsolete because schools, they do a good job of helping to create these learning communities that are incredibly, incredibly valuable. You look at the longer tail of different colleges and universities and the armies of different professors and educators, there’s a lot that goes into upward mobility. But I do think that what you’ll see more and more is... We talked about the athlete analogy earlier. Well, you don’t need to just be an athlete from a software engineering standpoint. You could be really comfortable just learning different software platforms and tools and no-code, the no-code movement, without needing to be a software engineer. I talk with presidents where they ask us to create coding requirements across all majors. Liberal arts schools that are trying to rethink how they can make all positions more competitive. You look at design or marketing. There are so many different roles nowadays, that a couple years ago you didn’t really need to know much about technology or tools or software. Now, you don’t understand data, you don’t understand how to make decisions based on data, you don’t understand how to leverage different tools? For so many roles, you’re kind of dead. So I think you’re going to continue to have these specializations.

Fuller: Let’s play that out for a second. If technology evolves, how do you see it affecting CodePath’s offer and your ambitions, not just in terms of what you’re teaching but how you’re scaling?

Ellison: We love to be as valuable to CTOs, VPs, vice presidents of engineering, as possible. So there’s always going to be an area of CodePath that creates those pathways into those most technical, most competitive roles. But then as we are able to get to a larger and larger percentage of the overall market of just tech and computing young people, then we’ll start to move into tech adjacent—from product management to sales engineering. As long as there’s high upward mobility, then it is potentially an area for us to examine and explore. We just want to make sure that, as best as possible, we arm our students for the most outsized economic opportunities. So we’ll continue to evolve. One of the greatest strengths of CodePath is, our curriculum is changing multiple times a semester. We open source our curriculum and our guide, so there’s professional engineers building alongside professors, building alongside these advanced students. Then we can be the bridge between education and what employers most want and most need.

Fuller: So, Michael, you’ve mentioned that the students aren’t paying for CodePath’s content or services. You said you had a nice liquidity event around a start-up, but how are you funding this thing?

Ellison: Yeah, good question. So not just students don’t pay, we pay them. We give students scholarships. We incentivize students from low-income backgrounds to be able to participate in our program, because it’s a luxury to not have to work multiple jobs to pay for books and other things. So there’s that. We don’t charge schools, we don’t charge professors. In some cases, we’re giving professors stipends. CodePath is supported by, we have quite a bit of earned revenue from major tech companies. If you’re able to deliver to the jobs and the roles that they care about the most, then it’s of high value to them. It’s at-scale talent matching, where we’re leveraging the data and the grading we’re doing across all the different schools to have a much more precise skills-based, role-specific matching of our students. That also allows our students to bypass stages in the interview process. So there’s a combination of these different offerings which also expand over time. Companies love to leverage CodePath to a brand and reach our audiences. We have funding from several venture capital firms—Y Combinator, Andreessen Horowitz as well—that want to connect with our students. So that’s been growing very quickly for us. So it’s over 100 percent year over year for the past couple years from that earned revenue. And then also we’ve received a lot of support from venture philanthropy organizations that care about driving a massive systems change in education.

Fuller: So it sounds like a blend of you’re getting some fees for placement, almost like a traditional job board. You’re getting some advertising revenue and also you’re getting philanthropic support?

Ellison: Well, I’d say the placement plus custom talent pipeline development, because we’re working at scale, we are working with underrepresented low-income populations, but we are designing the at-scale solutions to fill the jobs that you most want filled in your technology org. And we have a background redesigning, designing from scratch, global engineer onboarding programs. We created Airbnb’s first-ever engineer onboarding program, redesigned Meta’s global engineer onboarding program, redesigned Walmart’s global engineer onboarding program. So we’re leveraging our experience and our expertise with giving CTOs exactly what they most want, but doing that to give people from marginalized, underrepresented backgrounds the unfair advantage that we wish they had but they never do. And that’s our recipe for success there.

Fuller: So, Michael, how do you keep score? How do you think about results relative to your ambition of getting more diverse candidates, broad economic diversity, racial diversity, gender? How do you track your progress? And when you think about the type of magnitude of effect you’d like to have over time, how do you define that?

Ellison: Well, keeping score, if you’re focused on economic mobility, then you’re talking about where they end up. CodePath has over 10,000 alumni as of end of 2021. At the end of this year, it’ll be over 17,000 alumni, to show we grow pretty quickly. And if you stack rank all the employers, you look for the highest concentrations, the highest concentrations of CodePath alumni are in Amazon and Google and Microsoft in the hundreds. So that’s how we are holding ourselves accountable. We’re also looking at their upward mobility over time. Are their salaries increasing? Are their areas of responsibility increasing? We eventually want to start tracking the percentage of our alumni starting tech companies—ideally that are starting tech companies that really move the needle for them and for society as well. There’s a lot of money paid to the most competitive technical roles. And our hypothesis and our focus is in the next 10 years to diversify the nation’s most competitive entry-level software engineering roles. So that’s in the billions and billions of dollars in terms of economic mobility that you’re looking at. And that’s very much what our focus is.

Fuller: Well, Michael Ellison, Founder of CodePath, it’s been just a pleasure hearing about your success and your ambitions for the future.

Ellison: Well, it’s been such a pleasure. I always love connecting with you, the way that you’re thinking about and driving change in the future of work, so thanks so much for having me.

Fuller: We hope you enjoy the Managing the Future of Work podcast. If you haven’t already, please subscribe and rate the show wherever you get your podcasts. You can find out more about the Managing the Future of Work Project at our website hbs.edu/managingthefutureofwork. While you’re there, sign up for our newsletter.