| Needed Python Developer (Jupyter Notebooks) - 100% Remote - 6+ months Contract at Remote, Remote, USA |
| Email: [email protected] |
| Hello, Hope you are doing well. We have an immediate requirement with our client and your skills and expertise seem to perfectly match our client requirement. This is a Long-term Contract Position. Kindly review the job description below and send me your Updated Resume & Expected hourly rate. NOTE: Open for C2C (W2 preferred). Any Visa is fine Except (OPT, CPT) Python Developer candidates experience with Jupyter Notebook. Heroku would be a nice to have, but the need in-depth experience with Jupyter Notebooks. Title: Python Developer (Jupyter Notebooks) Location: 100% Remote Duration: 6+ months Contract Job Description: Current Project : Two opensource projects working on enhancing Jupyter Notebooks with Heroku. Technical Skills: Python developer that can work open source and finish Enhance the experience of running Jupyter Notebook on the Heroku platform. In-depth experience with Jupyter Notebook Interview Process: Virtual interview likely two rounds Overview: Jupyter Notebooks are an extremely popular tool in the Machine Learning and Data Science communities. They provide interactive environments with an easy-to-use code/test/observe loop for prototyping, exploring data, and sharing results. This project will help Heroku gain traction among the ML/AI community by creating & improving OSS software that allows users to run Jupyter Notebooks easily in single-user mode (via JupyterLab) or multi-user mode (via JuptyerHub). The goal is for this repo to be easy to start using and still full-featured enough to not outgrow quickly. Because we have a hiring freeze, were tentatively planning to hire a contract worker to explore & develop easy tools for developers to run Jupyter on Heroku. Below is our planned SOW (statement of work). Projects: JUPYTER NOTEBOOKS Test the current heroku/heroku-jupyter repo - flag any instructions that dont work or are unclear. Test various common notebook extensions (nbextensions) out to see if they work as expected. e.g. toc2, variable inspector, etc. Will likely need to add nbextensions to requirements.txt and maybe something like jupyter-contrib-nbextensions to the start_jupyter file. Document any that dont work as expected & theorize as to why. Test to see if there is a way to get apt-get install commands working after start up (maybe not), e.g. in a localized env Unclear if this is doable. Test if you can attach a standard Heroku add-on (e.g., logs, or access to a different Postgres database) to your running Jupyter app, and have that add-on work easily via a notebook. (For example, do the DB config variables show up in the notebook ENV variables available to the notebook instance) Example: Test if you can use S3 storage through the Bucketeer add-on for retrieving/storing data in S3. Known Issues from Internal Testing / Future Resolutions: The following items are known issues from our internal testing, or areas of uncertainty. Package won't install because it requires underlying other-things-apart-from-pip-package to be installed maybe it works w/ apt-get (needs to be done at buildtime unless there's a way around that), sometimes perhaps it won't Run out of memory / compute power modify the underlying dyno in app.json or via the heroku CLI. Need GPUs wait till AI dynos come out, until then no luck Data Access: Need to access data outside of heroku get permissions (e.g. AWS keys), expose in your app via configvars. Need to access data inside heroku attach that add-on to your app or expose in your app via configvars (need to confirm this works well). Need persistent s3-like object store connect to AWS or wait till heroku object store comes out Need to share with teams - use a good password, and/or add things like IP whitelisting to your app. Otherwise, need to upgrade to running JupyterHub for additional authentication & team options. Think about whether there is an easy way to - optionally - have persistent package installs (via !pip install not just requirements.txt) - not good from an engineering perspective, likely wanted in some research environments JUPYTER HUB: Modify the current heroku/heroku-jupyter repo to work w/ Jupyter Hub. (Think about if this should be the same or a separate repo.) Investigate and recommend- Think about how admin pages, etc, should be potentially viewable by only Heroku team admins Look into JupyterHub + OAuth (& similar features) - how can this combine well with Heroku Look into Github integration - how can this combine well with Heroku github integration Add potential to auto-shut off dynos / notebooks that are not in use (configurable) Docker images are uploadable to Jupyter Hub. Modify this feature so that OCI images are uploadable instead, if possible. May include setting a default OCI image for all notebooks, that is over-ridden by uploaded images. (Note that OCI images are compatible w/ docker, so we think this is a doable task.) [When heroku filestore has GAd]* test out a Hub file explorer feature Deliverables: Code checked into Herokus public GitHub account under the correct OSS license Getting started and Usage documentation Recommendations on future work to make the projects more usable by the community Stay Safe & Regards! Ravi Thakur Technical Recruiter LinkedIn: https://www.linkedin.com/in/ravi-thakur-2b1452248/ Omtech LLC, Apple Valley, MN Cell No: 857-341-2131 The BEST way to communicate is through an EMAIL. Keywords: artificial intelligence machine learning sthree database information technology wtwo Minnesota Needed Python Developer (Jupyter Notebooks) - 100% Remote - 6+ months Contract [email protected] |
| [email protected] View all |
| Tue Oct 15 00:39:00 UTC 2024 |