Spring 2025. I’m starting a PhD in Computer Science at SIUE.
Four months post-stage-4 diagnosis. Fourteen months post-math-masters defense. With uncertain time horizons and clear research priorities.
This isn’t a traditional PhD motivation. But nothing about this is traditional.
Why Now?
Stage 4 cancer with a 2-5 year median survival. Why start a 4-6 year program?
Because the research matters regardless of completion.
PhD gives me:
- Time for deep research
- Access to computational resources
- Legitimate space to work on hard problems
- Collaboration opportunities
- Freedom from commercial constraints
Whether I finish or not, the work will exist. Papers, code, ideas, tools.
That’s the goal.
Why Spring 2025 (Not Fall 2024)
I actually tried to start in Fall 2024—right after the stage 4 diagnosis.
The timeline went like this:
September 2024: Stage 4 diagnosis confirmed, started chemo September/October 2024: Tried to get PhD admission for Fall 2024 semester
Problem: I needed GRE scores. Had an appointment scheduled. While driving to the test center, they canceled on me.
Rescheduled. By the time I got scores back and submitted paperwork, the deadline had passed.
At the time, it felt like a frustrating setback.
In retrospect, it was perfect timing.
Fall 2024: The chemo regimen they put me on was way too intense for academic work. I was barely functional. No way I could have handled coursework.
Winter 2024-2025: Treatment stabilized. Energy returned. Brain fog cleared enough to think.
Spring 2025: Actually ready to start.
Sometimes administrative delays save you from yourself.
The Research Direction
I’m focusing on:
Machine learning foundations: Understanding how these systems actually work, not just using them
AI safety and alignment: Making sure we build systems that minimize harm
Complex networks: Analyzing AI conversations and reasoning as navigable graphs
Statistical computing: Methods that bridge theory and large-scale data
All connected by a theme: building and analyzing intelligent systems responsibly.
The Stage 4 Context
Let’s be explicit about constraints:
Medical: Chemo cycles, scans every 3 months, unpredictable energy, surgeries as needed
Time: Unknown but likely finite. Optimize for meaningful work now.
Motivation: Cancer clarifies what matters. No time for bullshit.
Productivity: Expect 50-70% of healthy baseline. Plan accordingly.
This isn’t tragic. It’s just parameters for the optimization problem.
What I’m NOT Doing
I’m not:
- Collecting credentials
- Pursuing safe, incremental research
- Following traditional PhD timelines
- Caring about academic politics
- Pretending I have unlimited time
I’m:
- Working on problems that matter
- Publishing results as I get them
- Building tools others can continue
- Documenting reasoning thoroughly
- Optimizing for research quality over completion
The First Semester
Spring 2025 courses:
AI: Classical search (A*, DFS, BFS), MDPs, reinforcement learning, Bayesian networks—all unified under utility maximization
Research methods: Setting up the dissertation track
The AI course is perfect timing. I need to understand:
- How intelligent systems optimize
- What utility functions actually mean
- How learning and planning relate
- Where alignment problems emerge
The Bigger Picture
Two master’s degrees gave me foundations:
CS (2015): How to build systems - encrypted search, cryptography, distributed systems
Math/Stats (2023): How to reason rigorously - proofs, statistical inference, probability theory
PhD (2025-?): Combining both for AI research
This progression wasn’t planned. But it makes sense in retrospect.
Research Questions I Care About
Can we analyze AI reasoning as complex networks? Treating conversations and knowledge graphs as navigable structures
How do we ensure AI systems minimize suffering? Not just performance—actual ethical properties
What does alignment mean formally? Beyond buzzwords to mathematical specifications
Can we make LLMs more interpretable? Understanding what’s happening, not just what’s output
These aren’t all solvable in a PhD. But they’re worth working on.
The Time Calculation
Optimistic (5 years): Complete PhD, publish multiple papers, build lasting tools
Realistic (3 years): Significant research contributions, useful software, incomplete degree
Pessimistic (1 year): One good paper, open source tools, documented ideas
All of these are worth doing.
The PhD structure supports this work. The degree itself is secondary.
What Success Looks Like
Not: “Got the PhD”
Instead:
- Published research that advances the field
- Built tools that others use
- Left clear documentation for continuation
- Contributed to AI safety understanding
- Made my reasoning legible to future researchers
If I do these things, the PhD was worth it whether I finish or not.
Managing Expectations
I’m not going to:
- Finish in 4 years
- Publish in top venues (maybe, but not the goal)
- Win awards
- Become famous
I am going to:
- Do rigorous work
- Publish what I find
- Document thoroughly
- Build useful tools
- Engage seriously with hard problems
That’s enough.
The Cancer Paradox
Weirdly, cancer makes me more productive, not less.
Not because I’m racing against time (though I am).
Because it removes everything that doesn’t matter.
No more:
- Imposter syndrome (who cares?)
- Perfectionism (done > perfect)
- Fear of failure (worse things have happened)
- Academic posturing (no time for that)
Just: do meaningful work while you can.
What’s Different This Time
I did my CS master’s as a younger, healthier person. It was about learning and credentials.
I did my math master’s while managing cancer. It was about foundations and rigor.
Now, PhD with stage 4: it’s about making the work count.
Every project is: “Is this worth my remaining time?”
Most things fail that filter. A few pass.
Those few get everything I have.
First Steps
This semester:
- AI course: understand utility maximization frameworks
- Start complex networks research
- Set up infrastructure for dissertation
- Begin building analysis tools
- Document everything
The work begins now.
PhD Year 1, Spring 2025. Uncertain timeline. Clear priorities. Let’s do this.
Discussion