Implementation (2020 plan)
About this page
This is our tentative implementation plan, as of August 2020. It is always subject to revision, but is a somewhat controlled document.
About
Researcher Software Engineers provide specialized scientific computing and data management support to researchers, beyond what is currently offered by Science-IT. Their funding is guaranteed by departments/schools/other units, but after the ramp-up phase most funding is expected to come from the research projects themselves.
Services include, for example, software development, scaling up or optimizing computations, taking new technologies into use, and in general promoting best practices in new and existing research using computational methods.
Funding types and sources
Funding has three types:
Ramp-up/Guarantee (R/G): Ramp-up funding to do initial hires, until project funding takes over
Ramp-up: department/schools/other units allocate a certain amount of money to do hires.
Units which provide Ramp-up/guarantee get first priority for their projects.
Replaced with project funding (below), if there are no projects then used for basic services (below).
Project (P): External or group money, allocated by a PI for a specific task in their group.
Basic (B): Allocated from units for short-term basic service for all of its members.
Allows short, strategic assistance without microtransactions
Science-IT work is a type of basic work, but may be requested by the Science-IT team instead of the researchers themselves. (For example, Science-IT has a long list of inefficient hardware use and inefficient software practices which can keep RSEs occupied for a long time. RSEs can also work on Triton/scientific computing technical development projects, which helps RSEs gain competence for the rest of their tasks.)
Time allocation principles
We track time spent per unit. Fairshare algorithm: the unit with the largest “deficit” in time gets priority for upcoming projects.
Units which provide ramp-up/guarantee funding get priority for their projects.
Project funding replaces ramp-up/guarantee funding.
Time paid from basic funding is allocated to tasks within the unit with the greatest strategic benefit, for example helping an entire group to use better tools or fixing extreme waste of resources.
When a group provides project funding, they can decide the tasks the RSE will do.
Ramp-up plan
This is a rough estimate of the type of demand we expect:
Distribution of work |
|||||
|---|---|---|---|---|---|
2020 H2 |
2021 |
2022 |
2023 |
Long-term |
|
FTE |
2 |
2–3 |
3–4 |
3–5 |
4+ |
Project work |
20% |
50% |
60% |
70% |
70% |
Basic work for units |
50% |
40% |
30% |
20% |
20% |
Basic work for Science-IT |
30% |
10% |
10% |
10% |
10% |
Our initial survey reached only Triton users and had 40 responses, of them 60% said “quick consultation”, 60% said “short term, 1–2 days”, 40% said “medium term, weeks to months”.
Actual ramp-up depends on funding cycles, research timing, and human psychology.
Start-up funding (already guaranteed)
(section removed; to be placed elsewhere)
Funding practicalities
Principle: the daily rate is roughly equal to “senior postdoc/staff scientist” salary + overheads.
Principle: When working for a research project, the RSEs record those working hours in Halli to that project. The corresponding portion of the salary is then automatically charged to the project. Remaining hours are recorded to the Dean’s unit RSE project, and once a year we split these costs and send them to each department. [Updated 2020-11-05]
(details to be filled in by Finance)
Measurement and KPIs
Number of projects and researchers who have been given support
Number of researcher-days saved, as estimated by our customers.
Fraction of project funding vs total program funding
Communication
Units which fund us will be informed of our activities at least every 6 months.
“As open as possible, as closed as necessary”. All RSE program data, documents, and statistics will be public, excluding actual project funding and information from the customers.
Risks and ramp-down
Primary risk: making permanent hires, yet not being able to sustain the program long-term.
Mitigation: we will only hire RSEs which can be absorbed into Science-IT naturally should the need of this service fade away.
Risk: difficulty in reaching researchers and explaining what we do
Mitigation: Science-IT has a long list of researchers who are using research services inefficiently: they can be contacted directly to inform about this service. Helping them and producing best practice examples for the future can keep several people busy for years.
Risk: Researchers see need, but group leaders unwilling to pay
This is indeed a risk, but there is precedence from other countries that there are enough people willing to pay. There will likely be a slow start, but as time goes on, expenses incurred by this service can directly be written into the budget of funding applications.
In our ramp-down strategy, we absorb the RSEs into Science-IT, CS-IT as part of its development efforts, or into other existing teams.