Research Computing Teams Link Roundup, 8 May 2020
Hi!
Winter/spring semester is here, and our team is onboarding (completely virtually) new student interns. For the student working most closely with me, we’ve been very careful to spell out very specific goals for the first month or so, and scheduling lots of one-on-one chats with the whole team. So far it seems to be working well, but it’s definitely slower than the usual onboarding process! I’ll report back how it goes - are you bringing on interns (or new hires) now? What approaches have you been taking?
On to the roundup:
Managing Teams
How to Get Your Team to Challenge Your Ideas - Dave Bailey
We’ve talked before about the importance of having your team being comfortable to disagree with you and offer alternative suggestions. One thing I like here is two sets of suggestions, depending on whether you tend towards over- or under-assertiveness:
For the typically over-assertive
- Adopt the question reflex.
- Aim for balance in hearing everyone speak.
- Avoid generalization.
For the typically under-assertive
- Over-prepare.
- Learn some facilitation techniques/helpful phrass
- Be vulnerable.
Confusingly, I tend towards a bit of both — overassertive in group settings like meetings, and underassertive when speaking one-on-one with individual team members — so this breakdown that I can use contextually is helpful.
The Meetings and Rituals that Actually Make a Difference when Working Remotely - Sofia Quintero, EnjoyHQ
After two months of working remotely for many of us (at least in North America) it’s a good time to take stock and see if our new distributed team meetings are working well. Here’s a look at what one team does that’s been fully distributed since 2015:
- Check-ins (online text “stand-up” report): ~2min daily
- Standup: ~15 min team meeting daily (as much for the face time as the standup)
- One-on-ones: they do 1 hour individual meetings monthly, which IMHO isn’t frequent enough
- Retrospectives: 1 hour team meeting monthly - what went well/what went wrong
- Weekly whole-team emails from the manager
My team (like a lot of ours, I think) works quite independently on quite different pieces of work, so I don’t know if team-wide retrospectives would be useful, and for the same reason I’m not sure about the checkins. On the other hand, I do annual “Christmas letters” looking over the year as a whole, and they seem well-liked; I wonder if something more frequent (maybe monthly, rather than weekly?) would be worth thinking about.
Managing your own career
Stop Zoning Out in Zoom Meetings - Sarah Gershman, HBR
Without the social cues of being near other people, it’s hard (especially after weeks of this!) to stay focussed on an hour-long virtual meting. In this article, Gershman gives fairly concrete advice - just give yourself something to do as part of the meeting. Make a plan ahead of time to contribute in a concrete way. With that plan, it’s much easier to stay engaged. She makes five specific suggestions:
- Define your value beforehand
- Acknowledge previous statements [i.e. - be listening!]
- Connect the dots of others comments
- Bring your attention back
- Ask questions
Of course, if the meeting isn’t worth your focus and effort for that period, well then fair enough, but in that case should you even really be attending that meeting?
Growth - Fred Wilson
A VC describes growth he sees in several of his new technical leaders. He sees two different syndromes in them; it’s much easier for him to grow the “Deer in headlights” new leader as opposed to those suffering from the incorrect confidence of “I got this”.
Relatedly, if you find yourself managing a new team, this from Claire Lew is a good primer on getting ready for a first team meeting.
Research Software Development
Codespaces - GitHub
GitHub Introduces Codespaces, Discussions, and Extends Security Features - Sergio de Simone, InfoQ
There were some announcements at GitHub Satellite 2020 this week. Codespaces, a development environment within GitHub itself, with a lightweight in-browser Visual Studio Code or connectable remotely to your own VS Code, looks really cool.
The new Github Discussions looks a lot like Issues, but allows threaded Q&A and — this is something that people who need to prioritize feature requests have wanted for a while — upvoting. (Woohoo!)
Discussions look to be live on GitHub teams pages now and will be in beta soon for public discussions; Codespaces is in free beta (you can request an invite), not clear when GA is or when or pricing will be announced.
How I Do Performance Reviews for Software Developers – That Are (Hopefully) Fair - Gergely Orosz
A good article on how one technical manager does performance reviews for software developers, but there’s no quick wins here — the key is to have very clear expectations as to where the developer should be in their skills and what they should be accomplishing.
Complexity has to Live Somewhere - Fred Hebert
You can contain complexity and move it to where it causes the fewest problems, but inherent problem complexity can’t be refactored out of existance.
Research Computing Systems
New Study Exposes Hidden Costs of High-Performance Storage - HPCWire
Panasas commissioned a survey by Hyperion on HPC centres’ criterion for purchasing storage systems. The results are predictably grim; storage continues to be a weak spot for most research computing clusters, with routine downtimes and difficulty hiring staff. Maybe that’s partly because the criteria people mostly use for buying the systems are performance and cost to the operator. I think this is one area where the needs of systems operators and those of researchers aren’t very well aligned…
Penguin Computing to Virtually Showcase and Demo at OCP Global Summit 2020 - HPCWir
Interesting to see long-time HPC-on-demand service provider Penguin demoing Open Compute systems at the upcoming summit. OCP is an effort largely started by hyperscalers to further commoditize servers by defining standards for servers (“compute modules”), hardware management, firmware, PDUs. This could be very helpful if it continues to propagate into research computing environments more generally. Has anyone else started to see OCP systems routinely showing up in RFPs for procurements?
Product Management and Working with Research Communities
A graduate student perspective on overcoming barriers to interacting with open-source software - Oihane Cereceda, Danielle E.A. Quinn
It’s easy to forget how confusing and intimidating it can be to work with open source projects for the first time - filing an issue, submitting a PR (is this change too trivial? Am I submitting the PR right?). This is a description from the point of view of a grad student on the issues with interacting with open source communities for the first time
Emerging Data & Infrastructure Tools
PostgreSQL, Oracle … graph query language standards adoption begins - Alastair Green
One of the reasons graph databases have been a little slow to take off in research computing (IMHO) is lack of a standard - sure there’s well established tools like Neo4j but what if that goes away?
In 2021 the SQL standard is widely expected to add SQL/PGQ, for defining property graphs in SQL; this is already implemented in PostgreSQL and Oracle. From there, GQL proposes a Neo4j-like graph-specific query language that makes use of these property graphs.
No-cost online AWS training pathway for researchers and research IT - AWS
AWS has posted a research-computing specific pathway through some AWS training material (6hr 55min worth of material plus 7hr 30min of additional optional material). Interesting to me (and the right approach IMHO) is they don’t go the route others take of introducing VMs-as-servers and then use ParallelCluster to emulate an on-prem cluster; instead, on the compute side they go right into containers, AWS Batch, and Kubernetes. On the storage side they cover FSx for Lustre, EFS, and S3.
Calls for Papers/Proposals/etc
VMworld Call for Speakers - Deadline 15 May
VMworld is a conference series with some relevance to many of us but likely wouldn’t schedule travel for; with a virtual VMWorld this year, and an AI, ML & HPC Track, it might be worth attending. If you have anything to talk about involving VMs, Kubernetes, microservices etc. (doesn’t have to be VMware specific), it might be worth submitting a speaker proposal. Speaker notification is 23 June.
Events: Conferences, Training
ContainerCamp 2020 - May 22 - $25/$99
A day worth of talks about containers and services.
Postgres Vision - 23-24 June, Free
I bet that at least one of the software stacks you support or work with involves PostgreSQL; with this year’s conference free and online, it’s worth taking a look at once the schedule comes up.
Random
A quantum computing simulator for laptops or servers, that also works well with the cloud providers’ quantum computing systems, for if any of your reseachers or team members want to start playing around with small quantum circuits.
Yes I know, we’ve all already got our favourite Zoom backgrounds now, but are you sure you don’t want one from the BBC: a Dr Who’s TARDIS one for meeting with some technical staff; and maybe a Yes, Minister background for meeting with the VPR?
A nice description of database consistency models brought to you by the team that very much tries to break database consistency.
Good tutorial on SSH port forwarding for those of us - ok, me - who always has to think for a minute about whether to use -L or -R for a given use case.
Oh and speaking of SSH, OpenSSH 8.2 now “just works” with U2F/FIDO2 keys.
That’s it…
And that’s it for another week.
Have a great weekend, and good luck in the coming week with your research computing team,
Jonathan
Jobs Leading Research Computing Teams
Manager, Research Science - AWS, Seattle WA USA
We’re looking for a Manager, Research Science who combines exceptional technical, research and analytical capabilities to build and lead a team that will be integral to the continued business growth of the Devices Organization. As a Research Manager, you will be responsible for leading a team of researchers and data experts in the design, development, testing, and deployment models
Manager, AI Research Engineering - Pearson, Denver CO USA
The AI Research Engineering team is responsible for all backend work supporting the R&D of AI products, services and prototypes. Our team’s development stack includes tools and technologies like Python, Google Cloud, AWS, GitLab, TensorFlow, SQL, REST APIs, Docker, and Kubernetes.
Program Manager, Research - Deepmid, London UK
Organise and optimise research into clear goals, objectives, sub-projects; Promote the clear articulation of team(s)/project goals and breakthroughs
Data Science Manager - Decode_M, New York NY USA
We’re actively looking to recruit Research Managers with 3-6 years of experience who, among other things, can fearlessly drive client projects from start to finish. This position, based in New York City, requires solid quantitative and qualitative research experience, a take-ownership attitude, and a hunger to learn and grow while delivering results that exceed client expectations – and have a blast doing it.
Senior HPC Engineer - Perkin Elmer, Cambridge UK
This role would be responsible for being a key member of a small team that is developing High Performance Computing Solutions, as well as Data Lake Platform solutions. The right candidate would have extensive experience in Linux scripting, HPC build and configuration, Data Lake technologies, as well as familiarity with cloud implementation.
Associate Dean for Discovery, Access, and Technology - James Madison University Libraries, Harrisonburg VA USA
In this role, the AD for DATech will provide vision and leadership for library technology infrastructure and collections, including physical, digitized, born-digital, and electronic collections. This division of the Libraries includes the following areas: Library Technology (including Learning Technology Services, Web and Application Services, Systems and Operations, and Desktop Support); Scholarly Resources & Discovery (including Collections Analytics and Resource Management, Access, and Acquisitions); Metadata Strategies; Digital Collections; Physical Collections; and Interlibrary Loan and Course Reserves.
Director Project Management - Sophia Genetics, Lausanne CH
Your key responsibilities will be to set project management standards and best practices across our company, to lead detailed project management efforts in our Information Technology, and Data Science organizations, and to oversee the successful releases of our portfolio of products. We have several key products with a global user base, and an ambitious and exciting Product Roadmap with complex projects that span many disciplines, such as scientific research, data science, software engineering, machine learning, medical device development, and operations management.
Founding Director, IT Research Computing Services - University of Rhode Island, Kingston RI USA
As the Founding Director, IT Research Computing Services the successful candidate will design, develop and lead IT Research Computing Services as the University continues its upward trajectory of engaging in higher levels of research across the University. The successful candidate is highly collaborative, interdisciplinary and possesses both strategic and tactical abilities as they develop a centralized solution to support the diverse current and future needs and interests of all stakeholders.
Quantum Computing Manager - Bayer, San Antonio TX USA
The Bayer Technology Prospecting team has been working on Quantum computing and Quantum Technologies since 2016. Bayer is also a founding member of QuPharm, which brings together more than 15 of the biggest life sciences companies all interested in Quantum computing.Be responsible for the evaluation and realization of the potential of quantum computing for Bayer, in close cooperation with the computational science and R&D community at Bayer.