# HOP(20/03/20)

## New Faculty
- Matt (Lecturer)
- Berrak Sisman (Speech, Spoken Language Processing)

### Interviewing
- Vision for Healthcare (Stanford Prof)
- AI - Reinforcement Learning (NYU Prof)
- Hardware (NUS Prof) x 2 (Memory Management, Computer Engineering, Computational Structures Enhancement)
- Network on Chips (NUS Prof) (Expected to come this year)
- Lecturer for Data Mining module

## New Modules

### T6

Product Design Studio (2021)
- Extension of what you did in Term 2 (3.007).
- Hackathon-type challenge with a company.
- Build end-to-end product solutions.

### T5/7

Service Design Studio (2022) (App Software design)
- Same as Product Design Studio but for end-to-end services.

Advanced Algorithms & Discrete Structures (2022)
- Replaces Prob and Stats (moved to Term 3 in Freshmore).
- Is an elective (w.e.f. 2021, ISTD students start choosing 1 elective in Term 5 instead of 6).

### T8

Mobile Robotics (2021) (Malika)
- Not really the hardware engineering side, but more on the "brain aspect".
- Use off-the-shelf provided hardware-interfaces for the "robots".
- Will be what Prof Malika applied in Canada.

Information Retrieval (2021)
- Advanced data mining course.
- CDS (Computational Data Science, Term 6) is an entry-level data mining course.

Spoken Language Processing (2022) (Berrak)
- Speech recognition, etc.

## New Stuff

- I3 lab is almost ready. Once the server is set up and the computers arrive, then it can be opened to ISTD students for use for projects, etc.
- Students entering SUTD in 2020 can start taking electives in Term 5 (instead of Term 6).

## Issues raised

**Software Engineering (SWE)**
- Lack of focus track/modules for people interested in pursuing mods related to software engineering.
- Software track as a "safety net" for people who aren't execeptional at or particularly interested in the existing specialisations (ML, Cybersec, etc.).
- ESC could be modified to have a greater focus on SWE instead of the current style, then Design Studio could be an extension of ESC.

*Response*
- Unfortunately, SWE is not a priority of ISTD. HoP is more concerned with how average ISTD students can turn out to be more marketable and competitive than graduates from other, larger, universities, hence the focus on equipping the average student with "lucrative" skills like AI instead of producing good software engineers.
- Software vs hardware: priority on hardware (robotics, IOT) + other stuff like AI as it will make students more marketable. Focus is on specialising students from as early as possible.
- HoP is thinking of a suitable track to park the Design Studio mods.

*Follow up*
- Student Board will come back to HoP with several suggestions for a dedicated SWE track.
- Plans for design studio mods (collaboration with DAI) will be consulted with students again.

**Intro to Prob and Stats**
- Due to current Covid-19 situation, cancellation of lectures with no e-learning materials for Probs and Stats. Cannot follow textbook/MIT OCW.
- Because cohort sessions are now covering the lectures as well, there is a lack of time to go through all the cohort exercises.
- Seniors said that content spiked too fast at the end of the course for their year.

*Response*
- Lecture slides prepared based on MIT's, SUTD's pace is apparently faster than MIT.
- No constraints regarding elearning. Acknowledged that they should have started recording lectures earlier.
- Possible to have recordings for lecture videos. Will have recordings for lectures and cohort classes will be focused entirely on the cohort exercises.
- **As of March 23, recorded lectures have been uploaded to eDimension**
- Seniors' feedback have been taken into account and the course should have been modified accordingly already.

**Computer System Engineering**
- Lack of a clear detailed curriculum objective.
- Need to know how students will be tested.

*Response*
- Will need to find a compromise between practice and given objectives.
- May release extra practice qns and stuff.

*Follow up*
- Due to mid-term survey, Prof Nat removed her summary notes from eDimension so as to prevent students from being obliged to read and memorise all her notes. Her notes are still available on request.
- Prof Nat mentioned that the take-home quizzes will be longer so as to give us more practice.

**Network Security**
- Prof cannot communicate. There is too much of a communication barrier.
- No access to the textbook which the instructor is referring to.
- Lab content is not good.
- Papers are not being marked properly.

*Response*
- The prof will not teach Network Security anymore. NetSec will be taught by another prof in the future.

**Deep Learning**
- The way the prof acts (carries the mod) is quite aggressive/hostile.
- Difficulty of the module is not an issue, but may scare people off.
- Course is good and well-designed.

*Response*
- This will be pointed out to the prof.

**Intro to Algorithms**
- Profs cannot deliver content effectively. Feels like they're just reading off the slides.
- More in-class exercises/actual implementation of the algorithms.
- Both Prof Ernest and Prof Ioannis needs improvement in how they deliver their content. Students prefer the "Oka-style" of teaching similar to Digital World.
- Suggested to use online judges such as HackerRank/LeetCode as practical homework.

*Response*
- Will look into using online judges as homework
- Will try to integrate integrate a greater amount of actual practical practices

## Action items

### Network Security
- Enforce a check of the final paper - both setting and the marking

### Intro to Algorithms
- Possible online judges (Code practice platforms)
- Leetcode
- Kattis
- HackerRank
- Might be able to use some of the beginner questions as homework to be solved by students

### Modern SWE skills ISTD is lacking

#### Software Architecture Design

- Dependency Injection
- Advanced Design Patterns
- Infosys touches on the concept but not very deep
- Asynchronous Programming

#### Software Engineering

- Test Driven Development
- What makes a good test?
- How to make testing part of the development process.
- Elements of Software Construction course is strangely obssesed with the definitions of obscure terms like "condition based testing" or manually calculating metrics that testing frameworks will generate for you ("number of paths executed"). We need to learn **how to utilize** the testing frameworks well, not how to become the testing framework!
- Orchestrating tests: how to setup local test environments depending on the project. Good time to introduce VM and Container orchestration tools (Vagrant and docker-compose respectively)
- Full Stack Development
- Dont need to teach one particular framework, but the concepts common to all frameworks
- Routing, middleware, MVC
- HTTP, REST APIs
- Database schema design (see next point)

- Databases
- **RDBMS (relational databases) should be its own mod.**
- The study of RDBMS concepts and schema design is a fundamental skill of full stack development. Shoe-horning big data technologies in the same course hampers this. (Senior feedback)
- Traditional RDBMS and Big Data technologies are **very different technologies** with **very different use cases**. We should go beyond iterating the benefits of big data technologies.
- Feedback from seniors said that the big data portion of the course (second half) felt like a giant info dump of Big Data technologies

- Build Tools and Processes
- It's 2020, and we're still including `.jar` files of software libraries in classpath manually... (Elements of Software Construction)
- For Java: Maven, Gradle, etc
- Again, not mastery of one particular tool, but the concept behind them. Why do we need a dependency manager? The same concept applies to other languages and franeworks, like `npm` for NodeJS.
- Basic understanding of toolchains like dependency management and build tools are probably the "basic things SUTD students dont know" that NUS SoC students are referring to.

#### DevOps and SysOps

Another skill highly in demand in industry that is not necessarily AI/ML. Requires intermediate to advanced knowledge of cloud providers like AWS, GCP and Azure.

- Basic Cloud and Networking Concepts
- VPS (Virtual Private Server), VPC, Subnets....
- Serverless workers (lambda functions)
- Security Groups
- High scalability and reliability
- Multi-AZ deployments
- Auto-scaling groups
- Continuous Integration and Delivery
- How to utilise build pipelines and deployment agents
- Blue/green deployments

Points above are just examples. Alot more information and training materials can be found from the respective cloud providers' websites.