Using Industry Standard Tools in Computer Science Education Neil Ernst Assistant Professor Computer Science University of Victoria [email protected] 

About Me Software architecture consultant, CMU Currently assistant professor in CS at UVic Undergrad in Geography/Geomatics PhD in software engineering and CS Learner Educator Study software design and use in the “machine learning” era Extensively study and use data science in my research Researcher Learner 2 

The Challenge: Industry Needs “15+ years of Tensorflow” Analytical communicators: writing and speaking collaboration and teamwork Technical skills: optimization, discrete math, stats data structures, algorithms, neural networks 3 

The Challenge: Academia’s Role Courtesy Jessica Shabatura 4 

The Challenge: Pragmatics • In an engineering discipline, → Show real problems, in 13 weeks, 3 hours a week → About one full work week! • From instructors who are – (mostly) Not practicing engineers – Constantly working to keep current (a security course from 2015 is now outdated) “Ancient and shriveled, many people said he hadn’t noticed he was dead. He had simply got up to teach one day and left his body behind him in an armchair in front of the staff room fire; his routine had not varied in the slightest since. Prof. Binns: 5 

Solutions • Use industry-like tools (for cheap) • Leverage groupware+collab software • Study real projects 6 

Industry Tools: Github • How to work in teams (version control, task management, planning) • Previously: – expensive enterprise licences (Perforce, TFS) – academic-only tools (Moodle, Blackboard etc) • Github (and others) offers industry-standard tooling for collaboration and versioning 7 

8 

9 

10 

11 

Industry Tools: Github • Free to use • Similar or identical to tools used in industry • Teaches concepts of version control by repeated practice • Not just for CS courses! • Other free alternatives: Bitbucket, Gitlab 12 

Industry Tools: BetterCodeHub • Problem: motivate need for code quality • Solution: BetterCodeHub leaderboard 13 

Industry Tools: Slack • Other groupware tools like Piazza very useful, but no use in industry • Force students to standardize on collaboration • Potential pain for instructors – Requests come at all hours, possibly to your personal device – Overtakes other Slack workspaces (e.g. for research group) – Lack of clarity around nationals of “embargo’d” countries • Alternative: Gitter, IRC 16 

Study Real Projects: UVicDSA • Challenge: some estimate 60% of time as developer is reading other people’s code. • Solution: introduce tools and techniques to practice reading software • Pick a project under active development, moderately large. Get a deep understanding. Write documentation of the system. • Examples: Tensorboard, RocketChat (group chat), IPython 17 

18 

UVic DSA 19 

Ongoing challenges • Diversifying and broadening the intake – Encouraging anyone to take CS – Leveling up in math without discouraging those with less previous math skill – Retrain and re-skill • 2nd entry program like UBC’s BCS or Uvic’s M.A.Sc in Data Science • Ethical thinking • Privacy and data collection • Bringing research/analytics into software engineering 20 

Increase Participation: HighTechU • real-world, technology-focused experiential learning opportunities for high school students • 7 week, weekend academy to build skills • 8 week paid internships • Work in collaboration with local high schools and local industry (SendWithUs, CoastCapital) 21 

No content

Diversity Gender Identity Grade 

No content

Challenges & Solutions • Use industry-like tools • Leverage groupware+collab software • Study real projects • Broaden the pool 25 Web: neilernst.net • Twitter: @neilernst