Goblins | https://goblinsapp.com | Founding Engineer #2 | Williamsburg, New York | ONSITE | $100-200k (+ generous equity)
In education, we have always traded off scale and quality. In a 30-student math class, half of kids are ready to accelerate while half are still working on their foundational skills, but teachers cannot be in 30 places at once, giving each student one-on-one support.
Until now. Imagine 30 teachers in every classroom—that's how Goblins feels. Students draw math on any device, and our AI figures out the "why" behind their confusion, giving instant feedback and building conceptual foundations. (Preview: https://youtu.be/SH9UomzBMUs)
Goblins launched to a private beta in March with 15 schools and 300 students. Since then we've built a waitlist of 2,300 teachers, and we now have paid school and district customers across the US and Mexico. We've also raised from angels and received a large grant from the Gates Foundation.
We are in the unique position of having a product that is both highly differentiated and demanded. Anyone who joins now will be getting well-priced equity. By the end of this school year, this company will be much larger.
Current technical problems we're working on:
→ Building knowledge graphs that adapt to student performance in real-time, routing them to the optimal next problem to work on
→ Extending our handwriting recognition from digital canvas to real paper through Chromebook cameras (think real-time OCR on math equations through low-quality webcams with poor lighting and limited CPU)
→ Creating interactive mathematical lessons (think 3blue1brown, but conversational) that remediate student misconceptions
You'll be engineer #2 on the team, joining Sawyer (CEO, former Head of Design at X1, acquired by Robinhood, and math teacher) and Alp (CTO, previously Stripe and Amazon).
We move fast, care deeply about craft, and obsess over delighting users. If you're interested, email sawyer@goblinsapp.com :)
Lot of negatively here but I think this sounds really cool.
I think I'm reading that the product sets a problem and talks the pupil through it. The pupil works on paper and the product recognises and grades thier written answer. It then uses an expert system to figure out the best concept to cover next, going back to explore the missing knowledge making the current problem impossible or advancing to the next step?
As an augmentation of teachers or a form of homework, this sounds really useful, but I'm not a teacher just a parent so who knows.
I'm not sure why the negativity is here either. They have 300 pilot users and a waitlist of thousands, so I don't think dissing the product idea makes any sense. The market has spoken, and they want it. Seems like a great opportunity to do something meaningful for the students and also to use the latest AI in an appropriate use case.
How, exactly, are you using current-gen AI to solve those problems? Does there exist at least one example of a student performing well (e.g., in math) from your program?
why does it look like you're trying to teach computation, which is step #2 of "math"
AI and calculators can easily do computation for us, why not take Conrad Wolfram's approach and teach step #1 which is to identify the problem and understand what computation is required to solve it
Computation is quite important to understanding a problem. Often a more complex problem requires the ability to do simple computation and to build a intuition about numbers. Especially when we are talking things like simple fractions or rearranging basic equations. When a kid hasn't gotten a good grasp of fractions, they struggle to see why we can multiply both sides of an equation by x/x and then move things around to simplify the problem. To them, it looks like a magical step that follows no rules. Only with enough computation are our brains treat 7/7 as x/x as dx/dx as 1.
(I know nothing about the GP post, so I can't comment anything about them; I'm only relaying my own experiences from tutoring kids who struggle in math largely because they offloaded too much simple computation to a tool.)
> Computation is quite important to understanding a problem
Your "problems" are just equations. 3x = 2 is not a problem.
Here's a problem: you and a friend have 15 dollars and would like to enjoy a day at the movies. Movie tickets cost $7.50. Will you have any money to spend on concessions?
The equation that you'd hope a child would produce is something like 7.5 x 2 = 15. And 15-15=0. Ultimately, no there's no money left over for concessions. That's the skill we need to teach. After that whether or not they know how to _compute_ 7.5 x 2 isn't a big deal. Give them a calculator.
It's crazy how indoctrinated we all are thinking equations are problems and teaching kids how to compute is "learning math".
3x = 2 is a problem. Being able to abstract real world problems into math and being able to take that math beyond any real world problems one is familiar with is a major component of math. Often the real world problems are not found until after the math has already been researched. For existing knowledge, trying to learn the concepts and the math at the same time is much more difficulty and often the science for people who know the math starts out simplified to the point of not being applicable to the real world (perfectly spherical cow in a world without resistance).
>After that whether or not they know how to _compute_ 7.5 x 2 isn't a big deal.
If you want them to have a job with math any more complicated than counting out the exact change the machine tells them, they are going to need to understand equations. Kids who can blindly plug integrals into a solver but have no understanding of how to solve it also have no ability to take a real world problem and build an applicable integral to plug into a solver.
Modern education does often fail at teaching kids how to apply the equations back to real world problems, but that seems to be an issue where such problems are inherently harder and education is being dumbed down with many stakeholders feeling it is unfair for kids who 'know' how to solve the equation to miss a question because they don't know how to construct the equation given the problem (specifically because of the metrics that schools are measured by being gamed, Goodhart's Law being what it is).
practicing computation is important in math education, I think Eastern Europeans and Asians often have an advantage due to working through 5-10K problems / year
> practicing computation is important in math education
no, 100% no on this -- we have computers that can compute for us, and we have since the 50s
the problem that this company is perpetuating is the idea that humans need to learn how to compute, we don't
we need to learn how to identify that a problem we're facing has a mathematical solution and how to translate the problem into an equation that a computer can solve for us
Not parent, but why/how is not always difficult; actual execution tends to be difficult. There is also a more cynical take, but I try not to engage that on HN.
In education, we have always traded off scale and quality. In a 30-student math class, half of kids are ready to accelerate while half are still working on their foundational skills, but teachers cannot be in 30 places at once, giving each student one-on-one support.
Until now. Imagine 30 teachers in every classroom—that's how Goblins feels. Students draw math on any device, and our AI figures out the "why" behind their confusion, giving instant feedback and building conceptual foundations. (Preview: https://youtu.be/SH9UomzBMUs)
Goblins launched to a private beta in March with 15 schools and 300 students. Since then we've built a waitlist of 2,300 teachers, and we now have paid school and district customers across the US and Mexico. We've also raised from angels and received a large grant from the Gates Foundation.
We are in the unique position of having a product that is both highly differentiated and demanded. Anyone who joins now will be getting well-priced equity. By the end of this school year, this company will be much larger.
Current technical problems we're working on:
→ Building knowledge graphs that adapt to student performance in real-time, routing them to the optimal next problem to work on
→ Extending our handwriting recognition from digital canvas to real paper through Chromebook cameras (think real-time OCR on math equations through low-quality webcams with poor lighting and limited CPU)
→ Creating interactive mathematical lessons (think 3blue1brown, but conversational) that remediate student misconceptions
You'll be engineer #2 on the team, joining Sawyer (CEO, former Head of Design at X1, acquired by Robinhood, and math teacher) and Alp (CTO, previously Stripe and Amazon).
We move fast, care deeply about craft, and obsess over delighting users. If you're interested, email sawyer@goblinsapp.com :)