Direct-Entry PhD
Timeline for First 3 Terms
- Fall:
- There may or may not be a TA duty.
note
Fall 2026: You will have a TA duty.
- Onboarding week (1 week before the semester starts, or 1st semester week).
- Reading Course on your research topic.
- 1 standard course (pick one):
- CS 9548: Foundations of Machine Learning.
- CS 9875: Theoretical Machine Learning.
note
Take CS 9548 if you're uncomfortable with the prerequisites listed here. You must be honest with yourself here since otherwise you will struggle in the future.
- Weekly reading group on Decision-Making Under Uncertainty.
- Weekly group meeting (rotating speakers) to showcase what you've been working on to the group, and receive feedback.
- Weekly one-on-one with Agustinus, personalized topic.
tip
The meeting density looks heavier than it is. Group meeting, reading group, and 1-on-1 are all feedback loops on the same research thread, not separate workstreams. Your actual work is the thread; the meetings are how you learn more about the thread or get feedback from different angles.
- There may or may not be a TA duty.
- Winter:
- TA duty.
- 2 standard courses:
- CS 9840: Probabilistic Generative AI.
- CS 9670: Reinforcement Learning.
tip
Both have final projects. Make sure that your projects are well-aligned with your own research topic. Otherwise, it won't be productive since you're aiming for a very good research state in late Summer.
- Weekly group meeting.
- Weekly one-on-one with Agustinus.
- Reading group is optional.
- Summer:
- No TA duty.
- No courses.
- Full throttle on research.
important
Aiming for a polished, advanced draft/submission by July (TMLR is a great venue for this). See below on why.
- Weekly reading group. Topics by popular vote.
- Weekly group meeting.
- Weekly one-on-one with Agustinus.
You should also attend research talks. They are a great platform to get inspiration and to network with other researchers (e.g., by asking questions). In-person talks' attendance is mandatory.
Transfer to PhD
The transfer process to PhD is initiated in the Summer (3rd term). You need to submit the following documents to Agustinus, in early–mid July:1
It's in your best interest to be timely with this. The sooner you are converted to a PhD, the more stipend you will get.
- Latest CV.
- Motivation letter (1 page).
- What exactly is your goal in the future? E.g., being a Prof, being an excellent research scientist, etc.?
- Why is a PhD the right thing to do, given your goal?
- Why you'll be a good PhD student.
- Summarize your progress so far.
- Research proposal for PhD (max 3 content pages).
- Similar to the OGS proposal, but on a longer timescale.
- Include planned timeline: which conferences each year you're targeting to submit what paper, when you want to do internships, when to start writing the thesis, etc.
- Polished, advanced paper draft/submission.
- MSc transcript.
The same writing advice applies. Esp., write draft as soon as possible and polish often.
Agustinus will then write a recommendation letter and attach the above, addressed to the Graduate Executive Committee. They will scrutinize your case. From there, the case will be brought up to SGPS, who will have the final say.
Your job is to make it easy for Agustinus to write a glowing recommendation letter. This starts from the very beginning—your first semester.
Following the timeline above, together with the How to Grad Apprentice guide, are keys.
Non-Conversion
In case you are not successful in converting to PhD, you will continue finishing up your MSc Thesis for 2 more terms. Agustinus' commitment as a supervisor doesn't change: you're still guaranteed the same level of supervision until you finish.
A non-conversion isn't a verdict on you as a researcher. It just means the PhD path isn't the right fit with me specifically, at this moment. MSc Thesis completion is a fine outcome; several strong researchers take this path and thrive elsewhere or go to PhD later.