We (https://illumidesk.com) support Julia notebooks (leveraging Jupyter Server Kernels using a variety of frontends, such as JupyterLab, VS Code, etc). The Robotics department (taught by Professor Jessy Grizzle at the University of Michigan) uses Julia extensively. Over the course of the last three years we have learned a few things about Julia when compared to Python and friends:
- Julia is faster at reading CSV's compared to using something like Pandas (presumably because it's multi-threaded and Python is single threaded)
- Julia does seem faster with read operations, but it's important to write code that declares types (myIntArray = Int[] vs myIntArray = []), this way Julia would read from a contiguous block of memory instead of reading an array of pointers in memory
- Julia's packages are limited to Python's equivalents
Our take is that Julia is amazing for ML/DL but Python is better for entry level data scientists. Once the data scientist gets up to speed with Python then Julia is a nice feather in the cap for more specialized use cases.
IllumiDesk is looking for a full time or part time Jupyter Lab developer to help develop and customize JupyterLab extensions.
If you have 10/15 hours a week available and have experience with JupyterLab extensions then feel free to reach out to hello at illumidesk.com for more information.
Thanks for reading and have a great rest of your day!
IllumiDesk is a small company...but has some very large customers in the Higher Education and Enterprise verticals. The team needs help from an
experienced and driven Development Manager/CTO. This person would help shape IllumiDesk's future by creating a team of engineers to lead the design,
implementation, and maintenance of IllumiDesk's software and infrastructure artifacts.
IllumiDesk's tech stack is (mostly) composed of:
- Python with Tornadoweb and Flask
- NodeJs/TypeScript with Express
- MongoDB and Postgres
- Jupyter and friends, such as jupyter_server, jupyter kernel gateway, among others
- React and HTLM5
- WebAssembly, particlarly Pyiodide
- Kubernetes
- AWS EKS and GKE
- IaC with Terraform and Ansible
We spare no expense in providing our engineers with the tools they need to ensure they are as productive as possible. This includes space to do deep work!
If you are an experienced Development Manager/CTO and are looking to get in on the ground floor of a unique, proven, and growing project then reach out to
hello at illumidesk.com for more information.
Thanks for considering us and good luck with your next adventure, regardless of where that takes you!
I had to respond on this one! I'm 47 and don't code "full time" anymore, but do spend at least 20 hours a week coding (mostly because I love it, partly because it's good as a business owner to look for better ways to cut costs and improve the UX at the same time). If you have some domain level expertise for a particular field and _also_ know how to code you will do very well indeed. You will do great if you know where your weaknesses are and work on them. The best coders I have ever worked with were folks that were humble and knew they didn't know $%&!. So the fact that you are willing to say you aren't great makes me believe you are better than you think.
My only advice (disclaimer not a sage here) is to continue to specialize in an areas that hasn't been commoditized yet and to work on the process, rather than the specific technology. You don't need to know all the "secrets" of a language to make you more productive. That will allow you to charge higher rates than for example commodity developer. Also, another great way to work on weaknesses is to teach others, learn by teaching as they say.
I'm sure you will do great things keep it up, love your attitude.
- Julia is faster at reading CSV's compared to using something like Pandas (presumably because it's multi-threaded and Python is single threaded) - Julia does seem faster with read operations, but it's important to write code that declares types (myIntArray = Int[] vs myIntArray = []), this way Julia would read from a contiguous block of memory instead of reading an array of pointers in memory - Julia's packages are limited to Python's equivalents
Our take is that Julia is amazing for ML/DL but Python is better for entry level data scientists. Once the data scientist gets up to speed with Python then Julia is a nice feather in the cap for more specialized use cases.