3 ms·

Scientific computing is still frequently populated with half baked apps or outdated apps. There is a big amount of progress to be made here and the problems are

Scientific computing is still frequently populated with half baked apps or outdated apps. There is a big amount of progress to be made here and the problems are not boring at all. I've developed programs for: Gathering and analyzing climate data at scale Sample collection robotics Low power data logging SDR and many more. The really cool thing is that you dont need a phd in CS to be effective and enable other scientists success. One question for you, eslaught: Do you have any specific requirements that would immediately disqualify a candidate?

eslaught 8y ago
> [...] to be effective and enable other scientists success. I would reframe this slightly: What are the fundamental open problems in computer science that prevent our existing tools from working for scientists more generally? E.g. are there constraints (like the need for high performance in certain applications) that break assumptions or tradeoffs made elsewhere in CS? Personally, I believe there are such constraints, and that's why techniques which may work for CS in other application areas don't always work so well in scientific computing---and conversely why some of the historical solutions in scientific computing look so backwards to the rest of the CS community. For us, it's not just about helping one particular group of scientists (though, obviously our work is heavily informed by working closely with scientists at the lab), it's about bringing about a sea change in the way science is done, by improving the tools we use it do it (be those programming languages, compilers, libraries, frameworks, visualization techniques, AI/ML, or whatever). > Do you have any specific requirements that would immediately disqualify a candidate? We're new enough that I'm not sure we've got a crisp answer to this, but if you mail me I can ask around if there are specific points you have concerns about.

eslaught 8y ago

> [...] to be effective and enable other scientists success. I would reframe this slightly: What are the fundamental open problems in computer science that prevent our existing tools from working for scientists more generally? E.g. are there constraints (like the need for high performance in certain applications) that break assumptions or tradeoffs made elsewhere in CS? Personally, I believe there are such constraints, and that's why techniques which may work for CS in other application areas don't always work so well in scientific computing---and conversely why some of the historical solutions in scientific computing look so backwards to the rest of the CS community. For us, it's not just about helping one particular group of scientists (though, obviously our work is heavily informed by working closely with scientists at the lab), it's about bringing about a sea change in the way science is done, by improving the tools we use it do it (be those programming languages, compilers, libraries, frameworks, visualization techniques, AI/ML, or whatever). > Do you have any specific requirements that would immediately disqualify a candidate? We're new enough that I'm not sure we've got a crisp answer to this, but if you mail me I can ask around if there are specific points you have concerns about.