RA and PhD Research Path: Building a Long-Term Research Bridge
I think about RA and PhD opportunities as research bridges.
The value is not only the title. The real value is whether the position helps create:
- stronger research training;
- closer interaction with serious researchers;
- better project taste;
- clearer publication direction;
- reusable technical artifacts;
- stronger long-term research identity.
For my current direction, the question is simple:
1
Can this path increase my probability of becoming a stronger AI researcher?
Why RA Matters
A good RA position can be a powerful bridge because it puts daily work closer to research production.
It can provide:
- exposure to a real lab or research team;
- discipline around reading papers and reproducing results;
- feedback from senior researchers;
- experience with experiments, baselines, ablations, and writing;
- clearer evidence for future PhD applications;
- a chance to turn vague interests into concrete research output.
The most important part is not the label. The most important part is the output loop:
1
2
3
4
5
6
7
8
paper reading
-> research question
-> experiment
-> result
-> analysis
-> writing
-> feedback
-> next iteration
This is the loop I want to strengthen.
Why PhD Matters
A PhD is not only a credential. It is a long-term training environment for independent research.
For me, a strong PhD path should help with:
- developing research taste;
- building original ideas;
- learning how to judge novelty;
- writing papers that can survive peer review;
- contributing to open-source research infrastructure;
- connecting AI, quantitative finance, and real-world decision systems.
The goal is not simply to enter a program. The goal is to become the kind of researcher who can consistently produce useful, testable, and publishable work.
How I Evaluate A Research Opportunity
I use a few questions:
- Does it connect to AI agents, RAG, AI for finance, research automation, or quant research?
- Can I produce public-safe technical artifacts from the work?
- Will I learn from people with stronger research taste?
- Can the work lead to papers, demos, benchmarks, or open-source contributions?
- Does it strengthen my long-term research narrative?
This turns RA / PhD planning from a vague career choice into a research-system design problem.
Connection To My Current Projects
My current projects are not separate from this path. They are preparation.
1
2
3
4
5
6
7
8
9
10
11
PM2.0
-> memory, capture, routing, execution discipline
Pengyi Paper Auto OS
-> paper-to-idea-to-experiment research production
FI-C-C OS
-> finance-domain engineering and FICC workflow trials
HKUDS / LLMQuant / X2Strategy study
-> AI infrastructure + finance domain + strategy execution loop
Together, these projects make my research direction more concrete.
Instead of saying only:
1
I am interested in AI for finance.
I want to show:
1
I am building systems that turn papers, market ideas, and domain knowledge into testable research artifacts.
Public Output Principle
The public portfolio should show:
- what I am learning;
- what systems I am building;
- what research questions I care about;
- what artifacts I can produce;
- how my projects connect to RA and PhD readiness.
It should not expose confidential material, personal operational details, or non-public constraints.
The right public narrative is:
1
2
3
4
research preparation
-> project output
-> technical writing
-> stronger research fit
Current Direction
The direction I want to keep building is:
1
2
3
4
5
AI research infrastructure
+ quantitative finance
+ research automation
+ human review
-> long-term AI scientist path
RA and PhD opportunities matter because they can provide a stronger environment for this path. My job is to keep turning learning into visible, credible, public-safe research artifacts.