# AaltoSciComp/AaltoRSE presentation to Data Agents

## Basics
- Science-IT: not SCI, not IT
- Who we are: 8 staff (2 RSEs, or "we are all RSEs")
- Q: Aren't we 3 RSEs, or do you exclude yourself Richard?
A: Yes, we are more than two. But if you include me, then we have to include others, and in effect we all are. Two specifically hired, full-time RSEs.

- History
- 200x: M-grid (cluster for materials science)
- ????-2021: Finnish Grid [and Cloud] Infrastructure: Academy infrastructure
- 2021-????: Finnish Computing Competence Infrastructure: emphasize the usage and support.
- We are Aalto and national infrastructure

## What is our goal?

~~5 years ago: Provide computing infrastructure~~

or

Now: Support computational research

## Aalto SciComp

"Aalto Scentific Computing" (ASC), officially known as Science-IT, is the main team.

- Organizationally Technical Services in the School of Science.

- Where does the funding come from?
- Hardware
- Academy of Finland (periodic FIRI (Finnish research infrastructure) grants)
- SCI
- U-level
- Staff
- Member departments (mainly CS, PHYS, NBE) Who is related to PHYS? OK -> Plausible PHYS data agents(?)
- SCI
- RSE projects/funders

- From which Aalto departments the most customers currently come from?

- Users:
- Types of users:

- What are the most common customer scenarios or projects where the ASC is currently helping?
- See below

## Aalto RSEs

A "Research software engineer" is a bridge between *Research* and *Technology* (not just software).

Engineering is about applying science to practical applications. In particular, we focus on the *integreity* of the output: does it use best practices, and in the context of open science, is it FAIR ?

- Where does the funding come from?
- 2020: We started with two initial RSEs as a seed:
- SCI (50%)
- CS, NBE, PHYS (50%)
- 2021: ITS digi pilot grant
- ongoing: Individual projects
- We have advertised RSEs thourghout Aalto (e.g. BIZ)
- Q: is this ok, can e.g. BIZ researchers expect RSE services?
A: Yes, definitely OK! Due to the ITS grant, but in general we try to help wherever possible.
- How does project funding work
- Within member units (SCI, whole Aalto with ITS grant):
- 1 month free, then project should fund out of its grants
- Other units:
- ???, small service for free
- Other universities
- According to contract research policies (= €€€€€)
- What are the most common customer scenarios or projects where the RSEs is currently helping?
- See below
- Could we have an example of a project done by RSEs for a customer?
- See below
- From which Aalto departments the most customers currently come from?
- CS probably dominates, then other SCI departments for most of the rest. Some across Aalto (includes at least ELEC, ENG, BIZ, CHEM).
- What are the cool things you are currently working on?
- See below
- What are the challenges?
- Getting our message out to researchers(, while at the same time...)
- ... Running out of time to do existing projects
- Cross-university collaboration (organization silos of funding)

## Most important internal projects and services

This is a tour of our most important projects.

### Triton
- Computer cluster
- Each week, we spend about 100 years of CPU time.
- "Ship of Theseus": continually upgraded (unlike most clusters that are fully recycled periodically)

### Data storage
- On Triton: Lustre
- 5 PB
- Very fast for big data and computations
- Not great for many small files
- Department Teamwork storage
- Managed by department IT (but in essence by Science-IT team, too)
- Smaller but backed up
- Data vision: data is the most important part and needs a vision of easy accessibility.

### scicomp-docs
- scicomp.aalto.fi
- FAIR: open-source, Github manageded, findable and reusable. Often high in web search rankings.
- Lots of information on Triton and more.
- Some duplicated info compared to aalto.fi: how should they be compared?
- License: CC-BY 4.0 for text and CC0 for code snippets

### Daily garage
- Zoom, everyday at 13:00
- Anyone may come and ask any question
- "Front desk" to many larger problems
- We direct other tricky issues here for discussion.
- Also internal networking/teaching/socialization time

### Accessible computing
- Working on one's own computer isn't enough (or FAIR enough)
- Clusters are traditionally hard to use
- We are trying hard to make computing *accessible*:
- Rethink cluster usability as much as we can. Design to be easy to use.
- Interactive tools, such as Jupyter or Open OnDemand
- Training and support where needed

### Teaching: Coderefinery, Kickstart, and other teaching
- These days, big gap between *complexity of tools* and *education*
- Academic vs practical education
- Cancelled "Computer as a tool" course
- Extensive course lineup for all levels of researchers and students
- Courses target to practical tools of researchers, not duplicating academic work.
- All course material is FAIR (Github, open source, collaborative)
- Teaching processes are also open: https://coderefinery.github.io/manuals (+ community teaching training)

## Other services
???

## Example RSE projects and ASC support

### ParallelFTDT
- Group had software tool, designer has left
- New researcher needed to use it but it wouldn't install
- We spent several days/weeks trying to get it installed
- Realize "hm, this is made internally. The software is the problem"
- Create RSE project, fix software, it is still used.

### Polar actiwatch data collection
- Researcher wanted to do a major study using wrist-worn Polar smart watches
- All data collected in one place
- RSEs write data collection platform
- And set up data pipeline to process and transfer to Triton in a usable form.

### Software instllation
- Most scientific software is written quite badly
- Common tools exist, but are often not used well by developers (or users don't know how to use them)

### Genomics data upload
- Researcher has issue with data upload to a repository
- Requires special program ("Aspera") which doesn't work on Aalto networks
- ITS wasn't helpful in un-blocking firewalls because of specifics of this program
- Make work-around within CS department
- In 2022, finally get a permanent solution in Aalto firewall.

### KVKL data management
- ENG research group has some very confidential data that is manually preprocessed
- We help to automate this processing

### Custom courses: Julia
- MS department group requested a course on Julia
- RSEs found existing CSC course, requested it to be open-licensed, then presented the course internally.

### Code efficiency improvements
- getting an existing, working code to run more efficiently either by moving to cluster usage, or by improving the code

### Some patterns in RSE projects
- Few that are creating large softwares from scratch
- Many in using tools and best practices in existing projects
- Many about data collection
- Optimizing or improving existing code
- Releasing and reusability.

### Most common garage support cases
- Can't install software
- Can't transfer data
- Unix problems using our tools

## Relationships
- ASC and RSEs?
- RSEs are a subteam within ASC: for all practical purposes, the same team
- ASC and ITS
- ASC isn't U-level. ITS provides many tools we use. We repackage and support others using ITS tools.
- ASC and RES/IES
- We work closely (most is centralized to Data Agents and related work now)
- ASC and other schools
- Other schools don't support or staff, but sometimes do our hardware. U-level funds us for all
- ASC and department ITs
- Overlap

## Open science in ASC
- Most tools developed open by default
- Always prefer to use open tools
- Cross-organization collaboration

## Challenges
- Increased complexity of computing makes a larger gap between best practices and what people do.
- Support of researchers and science outside of SCI
- Even being able to reach them. And should we reach them?
- Ensuring that researchers knows our place and value.

## Future

- Where do **you** think we should focus?
- Reaching out to new departments and groups outside SCI, I believe you're helpful there
- Stay flexible, stay available
-

- How can we (both Data Agents and RSE) manage to become the "go-to" point for any questions concerning our field? Essentially: How do we get more visibility for our services?
- get local contact points, people in deparments and research groups that support their own groups and use you as the expert resource, transferring difficult issues to RSEs
- It would be great fi essentially any new researcher gets a "If you have difficult questions with regard to this/that go there".