Fix the curriculums so I can justify restarting a new grad hiring pipeline in the US.
CS (along with ECE/EECS) degrees have been watering down their curriculum for a decade by reducing the amount of hardware, low level, and theory courses that remain requirements abroad.
Just take a look at the curriculum changes for the CSE major (course 6-3) at MIT in the 2025 [0] versus 2017-22 [1] versus pre-2017 [2] - there is a steady decrease in the amount of table stakes EE/CE content like circuits, signals, computer architecture, and OS dev (all of which are building blocks for Cybersecurity and ML) and an increased amount in math.
Nothing wrong with increasing the math content, but reducing the ECE content in a CSE major is bad given how tightly coupled software is with hardware. We are now at a point where an entire generation of CSE majors in America do not know what a series or parallel circuit is.
And this trend has been happening at every program in the US over the past 10 years.
I CANNOT JUSTIFY building a new grad pipeline in cybersecurity, DevSecOps, CloudSec, MLOps, Infra Silicon Design, or ML Infra with people who don't understand how a jump register works, the difference between BPF and eBPF, or how to derive a restricted Boltzmann machine (for my ML researcher hires) - not because they need to know it on the job, but because it betrays a lack of fundamental knowledge.
I can find new grad candidates with a similar profile at a handful of domestic CS programs (Cal included), but (Cal specific) someone with a BA CS from LAS who never touched CS152, CS161, CS162, or CS168 isn't getting hired into the early career pipeline for a security startup when they took CS160, CS169L, or CS169A because they are "easier", or isn't getting hired as a junior MLE if they didn't take all the more theoretical undergrad ML classes at Cal like CS182, CS185, CS188, and CS189. And even worse if they are a BA DS without a second fundamental major like AMATH or IEOR.
> Give me a new grad with strong fundamentals, a love of programming, and an interest in the domain and I'll teach them in sixth months whatever they missed in college that's relevant to the job
I 100% agree. A lot of core foundational classes that at the very least build the mindset of how to problem solve are not offered or have severely reduced the curriculum and content offered.
> until the implication that it's learning the nitty-gritty details that's important.
Not what I meant. What I mean is you can't understand or ramp up on (eg.) eBPF without understanding how the Linux Kernel, syscalls, and registries work. If you don't have the foundations down, I can't justify spending $120k base plus 30% in benefits and taxes hiring you out of college.
> These are kind oddly specific criteria
I'm giving random examples from individual portfolio companies
> Are those really things you think new grads need to know
This is the kind of curriculum a new grad from Cal (be they on F1 OPT or a citizen) are competing with when my portfolio companies have hired new grads.
There is a level of mathematical or hardware-software maturity that is built into top programs abroad that make it hard to justify hiring new grads domestically.
In Israel, India, much of Eastern Europe, and China - all universities follow the same curriculum as defined by their Ministries of Education.
I can find new grad candidates with a similar profile at a handful of domestic CS programs (Cal included), but someone with a BA CS from LAS who never touched CS152, CS161, CS162, or CS168 isn't getting hired into the pipeline for a cybersecurity vendor when they took CS160, CS169L, or CS169A because they are "easier".
Your ideal candidate needs to know: assembly, trivia about the history of eBPF, an obscure data structure specific to a certain field. As a bonus you would like them to know a little electrical engineering, and written on OS kernel as well?
Are those really things you think new grads need to know? I'm not sure you could find more than a handful of mid level or senior engineers with familiar with more than 2/3 of that.
This reads to me like things you know, that you think everyone should know.
Why not details on the network stack? Or database design and internals? Why not file systems specifically?
Those are much more relevant to the majority of modern development that the differences between bpf and epbf.
There might be advantages to increasing the amount of hardware and low level courses in the curriculum. But, I am pretty sure that is not the primary reason for young graduates not being to find jobs.
It is for us in the cybersecurity space and the fabless design space (eg. RISC-V SoC). My portfolio companies have moved all hiring to Israel, CEE, and India as a result.
The only people we might consider hiring in the US are veterans from cyber related MOSes because they come with the right learning mindset and have enough practical skills to ramp up if there are skill deficiencies.
I was going to agree with you until the implication that it's learning the nitty-gritty details that's important.
I can teach someone the details on the job. Give me a new grad with strong fundamentals, a love of programming, and an interest in the domain and I'll teach them in sixth months whatever they missed in college that's relevant to the job.
However I've noticed that the fundamentals are so watered down, even at top-tier schools, that young devs like that are harder and harder to find.
That’s a different group of skills than are required to build and maintain a CRUD app with complex regulatory requirements at 1m TPS running on a fleet of virtual hosts which are geographically remote from each other. And someone who is great at going deep on hardware may not be good at that role.
Asserting that the way it was done in the past is the best way will always get agreement from some, but the needs of industry change over time.
Send me their resumes to my username at protonmail.com then. I'll be the judge to decide whether or not they have the skills needed to justify a US hire.
I have most of the skills you list, and I am extremely confident I could fill in the gaps within weeks.
Given your presentation here, I'd like to know where you work so I don't accidently end up working for you, or with you. The elitism is at an 11. We're not special, we just have a skillset that isn't exclusively webdev. BFD.
Sadly, not enough CS exposure (or not communicated the right way).
At your stage, I'd recommend staying at your current job and doing an online PT masters like GT's OMSCS in order to bridge the CS gap, while also building the YoE to make the jump. In this market, landing where you are at right now is a feat unto itself.
I'm also not sure about your citizenship based on your background - most companies (especially early stage) are hesitant to hire and sponsor early career (despite what HNers say).
It's not like hardware and low-level jobs were booming either. If anything, universities have been adjusting to requirements of the market whether it's right or wrong.
This is the huge trend I’ve noticed on the last ten years. I too would love it if CS students studied more operating systems, networks, and computer architectures. Software engineering is very much an apprenticeship and we’re building real things. Few of us will dabble in academic domains but we all dabble in large complex stacks, often times distributed.
Learning computer architecture is fine and good and useful, but in a secondary fashion to help understand the main tool people in CS use. However, it is not the core of what makes computer science "computer science", or even the main thing that makes CS grads hire-able, despite the common misconception that computer science is about computers.
I would emphasize something along the lines of an HtDP approach developed by Felleisen et al. which goes beyond just the coursework in the HtDP book [0]. It extends into several core courses and in fact much of the core CS curriculum was being overhauled by Felleisen until Northeastern unceremoniously decided to dumb down the curriculum to satisfy some idiotic administrative idea of "market fit" and the desire to homogenize content across their expanding network of satellite campuses. When the curriculum was implemented, companies became very hungry for NU CS graduates, esp. given their experience with them during co-ops.
CS curricula are sadly being bootcampified, because that is the will of university administration.
I hate to say this but it is not just the curriculums.
A lot of teachers are just plain out bad at teaching. For quite a lot its not their fault, they were taught flawed pedagogy and just blindly follow what they know like trained monkeys despite how ineffective it is.
If you've ever heard the phrase, "if you are not struggling, you are not learning", you know that quite a lot of people have been twisted by the beast that is education. Such tools usually follow and originate in actual torture techniques, but that is obscured purposefully to the unwary. Paulo Freire's pedagogy follows this.
There is no longer any place in academia for competency, or accountability. Its been destroyed and sieved for decades, and eventually there's no turning back. You hit a critical point. That unfortunately, in my view, is where we stand today.
All except the top 0.1% of the competent people were driven out through social harassment, the remainder eventually conformed to the lower demands because they made the pool of people who remained looked bad. There is no economic benefit to good teaching, and most teachers overall fail, and even the good ones fail too because education is a sieve process and poisoned students coming in may not overcome the adverse effects despite perfect effort and knowledge (which is rare). I'm not saying all do, but the vast majority with few exception, remain poisoning minds; and those are people who can't be fired and must be waited out to retirement.
The same thing happens with any government organization where you can't fire such chaff, and these teachers who are often unfit in the profession are who get to teach your children, and determine whether they become engineers or other productive members; that is unless you spend hundreds of thousands on a private school of repute (which are rare, and highly selective).
The dominant pedagogy has gone by many names, by-rote teaching, lying to children, common core... aimed at depriving people of the most basic skills needed to get to the end-point.
The system certainly won't be refining any Einsteins, it will be destroying them before they even consider picking up Differential Equations at 16. They won't have the background to even get interested because the curricula has been turned into a torture machine and their prospects poisoned before they knew it.
A few years ago we hit a critical threshold, mostly silently, for credibility that there will be economic benefit. School Administrations have done the unacceptable, moved goalposts, and done everything in their power to incentivize the 'forever' student.
Families today have watched those that have gotten those degrees (at around a 1-3% pass rate) unable to get jobs, they aren't seeing the economic returns, and the debt crushes those people. This is why there are fewer children today. The group of your most productive people are fallow.
CS has one of the highest unemployment rates at around 70% according to BLS data. ECE pass rate while somewhat better (but not much) requires mind-destroying classes, compressed into time that no working adult would be required to work (>40h/wk), and perfect recovery from the sophisticated torture techniques used which destroy any intuition.
Education today is more about hobbling the student with trauma so they'll remain a student forever, while maintaining the lie that they can have a better life if they complete something they do their best to prevent sometimes through quite arbitrary means.
Math is taught from Algebra in K-12 on, following a lying to children approach. The good teachers who buck the trend and excel competently are so rare you might find only 1 in a county, and they have not been rewarded; in fact they've been passed up for higher credentialed peers who couldn't teach.
There comes a point where you just have to gut everything and go back to the way things worked. A working system. For teaching that's a first-principled approach following the greeks, where every teacher must follow it without exception, with strict requirements to maintain those standards and psychological support for students who may have had such trauma imposed.
The last thing that should ever happen is an Algebra student getting to Trigonometry and immediately failing because they were taught a flawed version, and Geometry was passed, but not a 'true' Algebra, and by that point you can't go back because that Algebra professor burnt the bridges on a lag (without accountability).
The problems we face today are largely self-inflicted, through blind destructive people who won't stop unless someone else stops them, and they have removed any ability to stop them non-violently following Tolstoy's philosophy, and utilizing existing structure not unlike any other parasite/cancer that tries to kill its host.
Evil people are those that are blind to the consequences of their actions and continue regardless. History has a lot to say about how this impacts the fall of empires, and we will be living through a fall; and proper education has been purposefully withheld to create environments of people follow a complete compromise, succumbing to systems of control.
CS (along with ECE/EECS) degrees have been watering down their curriculum for a decade by reducing the amount of hardware, low level, and theory courses that remain requirements abroad.
Just take a look at the curriculum changes for the CSE major (course 6-3) at MIT in the 2025 [0] versus 2017-22 [1] versus pre-2017 [2] - there is a steady decrease in the amount of table stakes EE/CE content like circuits, signals, computer architecture, and OS dev (all of which are building blocks for Cybersecurity and ML) and an increased amount in math.
Nothing wrong with increasing the math content, but reducing the ECE content in a CSE major is bad given how tightly coupled software is with hardware. We are now at a point where an entire generation of CSE majors in America do not know what a series or parallel circuit is.
And this trend has been happening at every program in the US over the past 10 years.
I CANNOT JUSTIFY building a new grad pipeline in cybersecurity, DevSecOps, CloudSec, MLOps, Infra Silicon Design, or ML Infra with people who don't understand how a jump register works, the difference between BPF and eBPF, or how to derive a restricted Boltzmann machine (for my ML researcher hires) - not because they need to know it on the job, but because it betrays a lack of fundamental knowledge.
I can find new grad candidates with a similar profile at a handful of domestic CS programs (Cal included), but (Cal specific) someone with a BA CS from LAS who never touched CS152, CS161, CS162, or CS168 isn't getting hired into the early career pipeline for a security startup when they took CS160, CS169L, or CS169A because they are "easier", or isn't getting hired as a junior MLE if they didn't take all the more theoretical undergrad ML classes at Cal like CS182, CS185, CS188, and CS189. And even worse if they are a BA DS without a second fundamental major like AMATH or IEOR.
[0] - https://eecsis.mit.edu/degree_requirements.html#6-3_2025
[1] - https://eecsis.mit.edu/degree_requirements.html#6-3_2017
[2] - https://www.scribd.com/document/555216170/6-3-roadmap
-------------
Edit: can't reply so replying here
> Give me a new grad with strong fundamentals, a love of programming, and an interest in the domain and I'll teach them in sixth months whatever they missed in college that's relevant to the job
I 100% agree. A lot of core foundational classes that at the very least build the mindset of how to problem solve are not offered or have severely reduced the curriculum and content offered.
> until the implication that it's learning the nitty-gritty details that's important.
Not what I meant. What I mean is you can't understand or ramp up on (eg.) eBPF without understanding how the Linux Kernel, syscalls, and registries work. If you don't have the foundations down, I can't justify spending $120k base plus 30% in benefits and taxes hiring you out of college.
> These are kind oddly specific criteria
I'm giving random examples from individual portfolio companies
> Are those really things you think new grads need to know
This is the kind of curriculum a new grad from Cal (be they on F1 OPT or a citizen) are competing with when my portfolio companies have hired new grads.
TAU - https://exact-sciences.m.tau.ac.il/yedion/2021-22/computer_s...
IITD - https://www.cse.iitd.ac.in/academics/btech_links/curriculum....
Uniwersytet Warszawski - https://informatorects.uw.edu.pl/en/programmes-all/IN/S1-INF...
Babeş-Bolayai University - https://cci.ubbcluj.ro/wp-content/uploads/2022/04/Curricula-...
There is a level of mathematical or hardware-software maturity that is built into top programs abroad that make it hard to justify hiring new grads domestically.
In Israel, India, much of Eastern Europe, and China - all universities follow the same curriculum as defined by their Ministries of Education.
I can find new grad candidates with a similar profile at a handful of domestic CS programs (Cal included), but someone with a BA CS from LAS who never touched CS152, CS161, CS162, or CS168 isn't getting hired into the pipeline for a cybersecurity vendor when they took CS160, CS169L, or CS169A because they are "easier".