Recent content by rkr

I don't think I've ever encountered a quant with a PhD in finance. Most of us have backgrounds in physics, economics, CS or mathematics. In any case, it seems a significant part of your post is rooted in the assumption that your degree choice has predetermined how much you can earn, which I...

It is still possible even if you don't go to a feeder school. There's plenty of mid-tier firms that don't hire directly from feeder schools. To get to a top tier firm though, you will likely have to take a hop (e.g. first job, internships) at reputed firms in related fields (e.g. tech). As for...

This post generalizes the answer that you're looking for. I'll tighten the bound and estimate <=1% probability that a full stack engineering role for a web application will satisfy an interest in university or graduate-level mathematics, if the said company has 30 or more employees. My...

For context, I competed in the Olympiads and the Putnam. My coursework was on a theoretical track, e.g. my first introduction to statistics was Casella, which I got to after measure theory, and I skipped more practical coursework in probability and statistics. I now run a startup. Sampling from...

Answering OP, here's a few ideas for getting the foot in the door: - Go to your nearest startup incubator or meetup event and just tell everyone that you're interested in startups and willing to code for free. - Go to your university's job fair or any large conference (ICML, ACL, ACM, SC, PyCon...

No problem, good luck.

MATLAB is actually not terrifically efficient. You may get cheaper, faster mileage and good power efficiency by doing things with MATLAB on AWS, and only accessing your instances over ssh. You may also have access to university server compute resources that you're unaware of, and the ability to...

Context: I've hired dozens of programmers and conducted a few hundred technical interviews. The interview process is definitely imperfect. The way I see it is that there's 3 categories of things: known knowns (A), known unknowns (B), and unknown unknowns (C). We've definitely had situations...

I don't think you need to post code examples on GitHub to appeal to your employer. I've gone through thousands of applications and actually, on some 90% of the applications where I've seen it included, it has actually hurt the candidate instead. If you do go the route of including your GitHub...

Thanks for the tag. I think that this post answers your question. If you already have a set of firms you're applying to, they're in the best position to tell you. You'll have firms that prefer React to Angular, Ruby to Python, JVM (Scala, Java, Groovy) to C (C++), functional to imperative...

The most significant part of practical data science work is just {data munging, preprocessing, cleanup}. Kaggle unfortunately doesn't do a good job of teaching that, as it usually comes with a cleaned and standardized data set. Often the cleanup involves specific domain knowledge. Software...

I got into it by coincidence, really. We spawned off a class at a university. Wouldn't have been possible without the financial support of alumni who propped us up. The median work hours are about 10h per day. There's a lot more on-site work than remote work given the nature of IP protection...

Only two of us. The rest of us were in different disciplines: algebraic geometry, string theory, ergodic theory, homotopical algebra, experimental high energy, bioinformatics etc.

Absolutely. For starters, our hiring rate for the research job has been about 1 out of 3,700. The largest firm in our space has barely over 1,000 employees, of which really about 100 of them really matter for the core research. I estimate there's less than 10,000 jobs of my kind in the entire...

My advice is always to (1) be good at your craft, (2) set constraints, don't just make global maxima decisions. Regarding (1), take your claim about AI for example. I was at the most recent ICML and EMNLP, and my Fermi estimate's that we had about 250 attendees for every 1 company recruitment...