Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Launch HN: Edlyft (YC W20) – Paid Support Program for CS College Students
55 points by ehairston on April 7, 2020 | hide | past | favorite | 30 comments
Hi HN! We’re Erika and Arnelle, friends since high school and co-founders of Edlyft (https://www.edlyft.com). We help college students pass their most challenging computer science classes, by offering them group tutoring sessions, study groups, and guidance from peers who’ve done it before them.

Arnelle and I came into college as freshmen not knowing how to code, but wanting to graduate with a CS degree. We found the steep learning curve discouraging, the lack of support frustrating, and felt like everyone else was always ahead. Impostor syndrome hit hard. But we made it through! In the end, what made the difference was connecting with students and mentors who had come before us. My junior year out of desperation, I wrote a letter to a grad student who was willing to meet with me weekly to review concepts and connect me with other CS students. Without that support system, I probably would've been weeded out from the CS major and not here writing this launch today.

Despite almost being weeded out, Arnelle and I were fortunate enough to land internships and jobs at fine tech companies. However we kept thinking about all the talented people who could and should be succeeding and don't have access to the same tools that we did. If they'd had the same kind of support that we were able to create for ourselves, they could’ve not dropped CS and maybe pursued their dream job in tech. Finally we decided to quit our jobs to build the support program that can make this difference.

If you got into programming before college and/or grew up in an environment where you were encouraged to play with tech, it may be hard to appreciate what an enormous head start that is. For many people who didn't take that path before college, there's a huge culture shock in the beginning to learn the basics of computing. It's all too easy to get discouraged and think that you don't have what it takes, and the sink-or-swim culture of academia unfortunately encourages these outcomes. Just having access to someone who was once in your position and knows that you can do it--because they did it--can be a game changer, especially in STEM.

You might be wondering why universities don't provide this already. That's what we ourselves wondered while we were going through the experience. For a while, we were fighting within the departments to get more support implemented. But it turns out that the incentives just aren't there. Colleges mostly aren't incentivized to increase CS enrollment, as Professors want to focus on their research and budgets are tight. Instead schools cap the major and struggle to increase support as demand goes up. Students wait for hours in line at office hours to get help--sometimes as long as 6 hours. At Cal, almost half of students who take the intro CS class will not receive a qualifying grade for the major. For universities, this is just an attrition number, but we know that much of that so-called attrition is people who have every ability to succeed at the material but need the right kind of orientation and support. Given the incredible value and growing importance of CS in our economy, this is not a minor difference in outcome. This is a broken system that we’re determined to solve.

Once a student joins Edlyft, they are immediately connected to a group of students in their CS class and an older student mentor from their school who has aced the class before. We hire compassionate and patient student mentors who host weekly group tutoring sessions and on-call q&a hours. Every Edlyft student gains access to up to 6 more hours of CS help per week and becomes a part of a larger community of CS students. They answer each other’s questions over Slack, work together over Zoom, and rely on our growing school-specific playbooks that are kept up to date by the student mentors. This is the supportive ecosystem we wish we had.

We’re currently launched at UC Berkeley, UCLA, and UC Santa Cruz. But the vision does not stop there; We plan to expand to other schools beyond California as fast as is sustainable and see a clear path to expanding into all high growth STEM fields, like Data Science and Pre-Med. We charge a monthly subscription for students to join, and offer need-based financial aid to ensure Edlyft is accessible and inclusive. Our hope is that the students who succeed through this program will become mentors to the ones who are coming up later, and make some money in the process.

If you have any memories struggling with Computer Science, please share them below! Although many people on HN were programming from an early age, we also know that there many who came late to computing and have done very well for themselves. We’d love to hear your stories and share them with our students to remind them that it’s normal for CS to be challenging and they’re not the only one.

Most importantly, if you know a college student struggling through CS, encourage them to find a mentor who can guide them through. And put them in touch with us! We’d love to talk to them.

We’re excited to hear your thoughts on how we can make computer science better for college students!



I studied CS at Duke. If students in your program can ask the mentor questions about their problem sets, I highly doubt Duke CS department will allow this. Our program was hyperaware of cheating. Most problem sets were supposed to be done individually. They were using software to audit our submissions to catch cheating.

I think students would pay to get help on their problem sets. I knew many students that paid for Chegg. I don't know many students will pay for general tutoring.

Also, keep in mind, a lot of people try CS and don't end up liking it. Most people I know would not want to be a programmer even if they were guaranteed to get A+s in all their college CS classes. It takes a certain type of person to love staring at code on a computer screen 8 hours a day. It's not for everyone.


> If students in your program can ask the mentor questions about their problem sets, I highly doubt Duke CS department will allow this.

Came to write this. Homework help services were specifically called out in our course policies as a form of academic dishonesty. Getting outside help on a problem set would get you kicked out of the major. It would take a lot of care to design a tutoring program that was specific to our courses and didn't run afoul of our academic integrity rules.

OP: this is a nice service for universities that allow this sort of thing, but you should be sure you're not running afoul of programs with stricter academic integrity rules (which, AFAIK, includes most top programs).


I would like to echo the concerns about academic integrity. For example, at Cornell, it is not permitted to get help from anyone but course staff on assignments in the intro CS class (and many others): https://www.cs.cornell.edu/courses/cs1110/2020sp/policies/cs...

I guess this could still be useful for help with concepts/exam prep, but not sure how much traction that would gain as people tend to be busy with the projects.


> Most people I know would not want to be a programmer even if they were guaranteed to get A+s in all their college CS classes. It takes a certain type of person to love staring at code on a computer screen 8 hours a day. It's not for everyone.

I had a friend who graduated with a degree in "design" and got a job as a web designer. She'd tell me how the pay was good, but she just couldn't stand sitting at a computer all day.

She quit pretty soon afterward to be a teacher in an extracurricular enrichment program for very young children. The new company didn't even reliably make payroll... but she preferred it.


Out of curiosity, what software were they using to catch cheating?

Was it MOSS (https://theory.stanford.edu/~aiken/moss/)?


You know, it's really easy to get around MOSS. You can just run it on your own code and figure out when it gives a result that appears to be "not cheating". That number will depend on how much starter code your assignment involves.

Like seriously, MOSS just filters out the kids who are bad at cheating from the ones who are good at it.


Penn uses MOSS


Congratulations on launching. Getting my CS major was one of the hardest things I did. Not because the knowledge (it's hard, but not thaaat hard), but because of the perseverance you need to have in an enviroment as toxic as a college, with sometimes bad professor (and sometimes the best you could ask for), or bad colleagues / students (and also the best friends you could get).


Congrats on making it through and thank you for sharing! Your story resonates so much with me and so many of our students. Often the content isn't the most discouraging part, but instead the environment. So well said!


I'm surprised you launched at Cal given your world-view of incentive alignment. I was there right when the department got slammed. Clearly, it wasn't an ideal situation but there was a ton of innovation in the department to meet demand. Additionally, Cal makes it very clear that Profs are expected to teach, and many who teach lower division classes certainly make it their top priority when teaching (thinking of mainly of DeNero, Hug, Garcia, Fox but would also include Babak, Sahai, Vazarani in this group as well).

DeNero led a ton of software development to help 61A scale (fuzz testing projects, automated hints, autograding). Cal recognized that lot of people taking 61A were actually business or biology majors who wanted to know how to code for their own industries, so they added an entirely new major, complete with new courses and its own building, to relieve pressure on the CS department.

A now-standard company (Gradescope) was launched to help TAs grade faster, often returning graded midterms just a day later. Not to mention Piazza, which became standard in all CS classes within one or two semesters.

And not only that, but tons of tutoring resources were available for lower-division classes from other student groups like HKN/UPE (spoiler alert -- most of these tutors were current or former TAs so you got the same level of knowledge without the waits) and ASUC.

I'm sure a lot of people who've given blood, sweat, and tears to scale this program in a resource-dearth environment _because_ they love CS so much would be upset with your characterization of the department's stance.


Completely agree. CS departments deserve a ton of credit for having to adjust to insane levels of growth that are unlike any other department.

Also you're spot on about Cal. In fact, I too questioned whether or not it made sense for us to launch there at first, especially given that they have such a topnotch CS program with incredible resources. However when we spoke to 100s of students there before we launched, we quickly learned that the competitive environment still resulted in so many students slipping through the cracks. Moreover, we encourage Edlyft mentors to direct students to existing resources on campus as well. Edlyft should not be a replacement for these existing resources, but instead a supplemental option where students feel they can be a little more vulnerable.

Lastly, our characterization of the climate in schools comes directly from students' experiences and our own. Cal definitely has tons of resources that I only wished I had access to when I was in school. However many schools are still very far from getting there, leading to so much talent being overlooked. Edlyft should not have to exist and we're successful when it no longer needs to.


I think that the main reason that students want tutors and other resources is because office hours are slammed during projects, event though there are lots of TA's and readers. There are a lot of resources to help students with material in terms of understanding, but there are a lot fewer to help students debug projects, so I think that it's where a lot of the demand is. Unfortunately, I don't see Denero or Hilfinger allowing private debugging help for hire as a solution due to cheating and plagiarism risks. How do you ensure that the tutors are abiding within the honor code, and not helping students cheat?


Wow! You highlighted exactly what we've heard from sooo many students: OH being slammed during projects as a major pain point.

To avoid helping students cheat, we have all Edlyft mentors agree to an academic honesty policy, go through mentor training, and all are interviewed before they join Edlyft. We've also considered recording sessions as well, but haven't gone that route yet.

But more than that it's been fascinating to watch some of our mentor sessions. What we've seen happen is some students actually prefer to use Edlyft to become an "excellent" CS student and hone their understanding of concepts vs. getting hw done just for the grade. My hypothesis on this is that our mentor sessions are unlike traditional office hours where you may only get 10min of TA attention. So, when students are able to slow down and really grasp concepts, doing hw on their own gets easier.


"At schools like Berkeley, more than half of the students who attempt the intro computer science course will not meet the average grade required to actually declare the major. Without previous experience, it’s incredibly hard to make it on the right side of that curve."

I don't think this is true, and find this wording to be _more_ intimidating to potential CS students, as it gives the false pretense that it is "incredibly difficult" to pass CS classes with no prior experience.


So I think their claim is _technically_ true but leaves out a lot of context. I assume they're talking about CS 61A, which last fall enrolled _over 2000 students_[0]. That's almost 1/3 of the entire freshman class. The average grade for CS61A is actually a B+[1]. However, to declare as a CS major as a Liberal Arts major you need an average GPA of at least 3.4, which is higher than the B+ GPA (3.3). Thus, their claim is technically true.

[0] https://www.dailycal.org/2019/09/10/cs-61a-course-enrollment... [1] berkeleytime.com (doesn't allow deeplinking to specific page)


Actually the context is missing but in a kinda different way. Berkeley CS has about 2000-2200 students in 61A each semester, and an average grade of B+ (3.33). Students need a 3.33 average in 61A and two other classes, 61B and 70. However, most students don't take the other courses ("Only" 1750 and 750 this semester in each of those). Cal requires 61A for other majors too (Cognitive science, data science) so a lot of students take that class but don't take others. In essence, Im agreeing with you, but just being a little different semantically.


It is true because Berkeley intentionally does this in order to cap the number of people enrolled in their computer science major. Some people that even do well in their CS classes are not able to declare the major due to the high GPA requirement


Interesting idea. As a somewhat non-traditional CS student, I could definitely see this as a tool I would have turned to when I was starting out.

One of the most frustrating aspects of studying CS is that my professors would come up with problem sets and assignments on their own and their solution would require something not found in online resources. There were quite a few times where I could only find tangentially related information in very obscure places. Hopefully your service can address this issue effectively.

Also, it seems like the folks here on hn aren't your target demographic so the pain may not resonate here, but i think it exists in a deep way, none the less.


Glad our service could resonate with you. Thanks for support!


I came in to my undergrad institution (Georgia Tech) as a CS major but switched out after the first year for reasons similar to what you've described.

A lot of my classmates had been programming for a while before coming to Tech for their undergrad and compared to my peers I was having trouble keeping my head above water with my CS classes and was struggling to access support.

Sometimes I wonder if I should have just stuck to the program but in retrospect I didn't really have the resources/guidance/support (in more ways than just academically) at that time.

I wish you guys success!


Thank you so much for sharing!


I think the idea of building a community is valuable, and reduces the risk of disintermediation if you were just a tutor-matching service. You might want to look into the idea of separate pricing for the tutoring component vs just being part of a community.


That's a great idea! Especially now that students are alone and off campus, the community aspect has become even more important.


This sounds overly specific, there are lots of places to find private tutors, so building one aimed specifically at undergrad first 2 years in a specific field - isn’t it a bit too specific? I’m trying to count how many unicorns are there in the tutoring marketplace, and nothing comes to mind. I mean if there was a Tinder and you would be the tinder for pets, I get it, the main category is saturated so you find a niche, but what is the poster child of tutor finding? I used some app in the past and it tried to convince me to use the app to pay the tutor, it ended up just being a way for me to get their phone number, I pay them directly (tutor for my kids). I think you either go with care.com approach where you pay a fee, or mask sending emails / numbers, otherwise, tutors will figure they can cut the middleman and get paid directly once you start scaling and can’t hire them directly. I wish you all the best and lots of good ideas get a lot of “it won’t work” naysayers but I’m curious how you got YC to invest in this, what evidence you have that people will want to pay for it, and that they will use your service vs any of the other “find a tutor” places out there. Last question, how do you make sure it doesn’t turn into a “do my homework for me” thing like course hero etc.

All the best, I’m just curious on how the business of this will work, and how you convinced YC to believe in it too.


You're spot on; we were very intentional about starting niche in these intro weed out STEM courses because we recognized that there was a huge gap in the system and a "hair on fire" problem we could solve: 50% of STEM students switch majors & most who completed undergrad had no idea how much their major choice would impact their career. Dropping these intro classes often is the difference between graduating with a STEM degree & an in-demand job or struggling for a lot longer. This felt like a low-hanging fruit spot to make huge impact.

Some poster children for tutoring are: Chegg, Kaplan and Crimson Education. The way we avoid the "off network" issue (which I've definitely done myself before as well), is by offering group sessions. With this model, we're able to offer students way more help for much cheaper than paying an hourly 1:1 tutor fee. Edlyft Mentors also love tutoring for Edlyft because they are usually CS majors trying to build their resume and we are a YC-backed tech startup. To avoid "do my hw for me", we have all mentors agree to an academic honesty policy, go through mentor training, and all are interviewed before they join Edlyft.

Love the question on convincing YC! Our key insight was that 70% of coding bootcampers already have their bachelor's degree. This is way higher than most realize. Edlyft is tackling the Lambda School problem from a different angle that is much lower cost and higher margins, meaning we don't need to teach curriculum. Instead, we reach students when they are already in an environment for learning and have less responsibility (no kids or full time jobs usually). We give them the support they need to graduate the first time with the skills they would have paid an additional thousands of dollars for in the future.


Nice... great reply, I see why they accepted you guys. I assume now you have more tutors than students, but will you open it as a "gig economy" for tutors in the future? (e.g. join as a tutor). Sounds like something I'd be happy to volunteer doing, let alone get paid for as a side gig :)


I had a very similar startup idea, it was less based on a niche though but the rationale is the same. I have startup ideas about education quite often because my vision on it is quite strong. If you want to chat about education, drop me a line (email is in the profile).

I created doodledocs.com for fun also to aid education as a side project. It's a purely frontend P2P app. I have ideas on education quite often and sometimes I tend to prototype them out.


I am thrilled now to see such a service now exist! I am always want to learn CS myself, but always felt hard to dive into cuz I have 0 STEM background,I always think group learning is one the the best way to learn CS, let alone now it also provides us a mentor, really interesting and hope this service become popular, I will definitely try it in the future, bookmarked.


Very interesting. Have you had any pushback from school administrators who may be concerned with wealthy students gaining an unfair advantage? Although I guess this is no different than private tutoring, which most schools are okay with as far as I know.


Great question! We've built our model so that for every full price paying student, we can admit a need-based financial aid student.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: