🎓 PhD Survival Guide

“The PhD is a marathon, not a sprint. The goal is not to be the smartest person in the room — it is to still be in the room five years later.”

Graduate school in computer science is unlike any professional experience before it. Unlike undergraduate study, there is no syllabus for your dissertation, no weekly deadlines to keep you accountable, and no guarantee that your work will succeed. This guide aggregates hard-won wisdom from researchers who made it through.

1. The Mental Model Shift

The single biggest mistake new graduate students make is treating the PhD like an advanced undergraduate degree. It is not. A PhD is an apprenticeship in the creation of new knowledge.

What Changes on Day One

Undergraduate Mindset	Graduate Research Mindset
Consume and reproduce existing knowledge	Create knowledge that does not yet exist
Correct answers are known in advance	Problems may have no solution at all
Semester-length feedback cycles	Multi-year feedback cycles
Success is passing exams	Success is publishing and defending original ideas
Professor instructs	Advisor guides (sometimes)

The Impostor Syndrome Trap

Studies consistently show that 70% of PhD students experience impostor syndrome — the persistent belief that you are less capable than your peers and that your admission was a mistake. Strategies for managing this:

Normalize struggle: The papers you read represent polished final products. You never see the years of failed experiments and rejected submissions that preceded them.
Track your progress, not your peers: A weekly research log makes invisible progress visible.
Seek community: PhD student slack groups, lab meetups, and conference coffees are not optional socializing — they are mental health infrastructure.

2. Structuring Your Time

The PhD offers radical autonomy. Without structure, weeks disappear. With too much structure, you optimize for busyness instead of breakthroughs.

The Weekly Research Template

Monday    — Deep work block: literature review / writing / experiments
Tuesday   — Deep work block: coding / experiments / analysis
Wednesday — Meeting day: advisor, collaborators, lab group
Thursday  — Deep work block: paper writing / revision
Friday    — Admin, reading new papers, planning next week

[!TIP] Protect at least 2–3 hours of uninterrupted deep work every morning. Email, Slack, and meetings are attention destroyers. Batch them into the afternoon.

Time Tracking Tools

Toggl Track — Free time tracker. Log your research tasks by category to identify where your time actually goes vs. where you think it goes.
Notion — Build a personal research wiki with weekly goals, paper notes, and experiment logs.
Obsidian — Local-first Markdown-based knowledge management with bidirectional linking. Excellent for maintaining a Zettelkasten.

3. Managing Your Advisor Relationship

Your relationship with your advisor is the single most important variable in whether you enjoy your PhD and finish on time. See 02-choosing-advisor.md for selection criteria.

Communication Protocols

Send a pre-meeting agenda: 24 hours before every advisor meeting, send a bullet-point agenda with: (1) what you did since last meeting, (2) what blockers you hit, (3) what you plan to do next week.
Document decisions: After every meeting, send a brief email summary of decisions made. This protects both you and your advisor from memory drift.
Escalate early: If you are stuck for more than two weeks, escalate. Staying stuck in silence wastes your most finite resource: time.

When Things Go Wrong

If your advisor is unresponsive or unavailable, establish a secondary mentor from your committee early.
If the relationship becomes hostile, contact your department’s graduate ombudsperson immediately. This is a structural problem, not a personal failure.

4. The Research Process Reality

Research is not linear. The popular image of a researcher having a eureka moment followed by a clean experiment is fiction.

The Real Research Loop

Curiosity → Hypothesis → Experiment → Failure → Revised Hypothesis
     ↑                                                    |
     └────────────────────────────────────────────────────┘
                          (repeat ~200 times)
                                  ↓
                      Successful Experiment
                                  ↓
                    Paper Rejected × 2–4 times
                                  ↓
                          Published Paper

[!NOTE] A typical top-tier conference paper (e.g., NeurIPS, CVPR, SIGCOMM) represents 6–18 months of work and 2–5 rejection cycles before acceptance. Rejection is the norm, not the exception.

Failure Journaling

Keep a failure log — a private document listing every experiment that failed and why. This has two benefits:

You avoid repeating the same dead ends.
You see, in aggregate, how much ground you have covered.

5. Health and Sustainability

Physical Health Basics

Sleep: Chronic sleep deprivation degrades the complex reasoning and creativity that research demands. Protecting 7–8 hours is a research strategy, not a luxury.
Exercise: Even 30 minutes of moderate exercise, 3–4 times per week, measurably improves cognitive performance and mood regulation.
Nutrition: Keep healthy food accessible. Decision fatigue is real; if junk food is easier to access than real food, that is what you will eat at 11pm.

Mental Health Resources

PhD Balance — Community specifically for PhD student mental health.
Therapy for PhD Students — Most universities offer free sessions with licensed therapists for enrolled students.
Not Alone: The PhD Mental Health Podcast — Honest conversations about the psychological realities of doctoral study.

[!CAUTION] If you experience persistent depression, anxiety, or thoughts of self-harm, contact your university’s counseling services immediately. Academic success means nothing if you are not healthy enough to enjoy it.

6. Building Your Research Identity

By the time you defend, you should have a clear, specific answer to: “What are you the person in the world who knows most about?”

Niche down early: “Machine learning” is not a research identity. “Efficient inference for vision-language models on edge devices” is.
Own your Google Scholar profile: Keep it updated. Your citation count is your academic reputation made visible.
Be on Twitter/X and LinkedIn: The research community is active there. Follow your field’s leaders, engage with paper discussions, and share your work.

7. The Long Game

Year-by-Year Milestones (Typical 5-Year PhD)

Year	Primary Focus
Year 1	Coursework, lab rotation, literature review, finding your research direction
Year 2	First research project, first workshop/venue submission, qualifying exam prep
Year 3	Qualifying exam, first major conference paper, solidifying dissertation topic
Year 4	Core dissertation chapters, internship (optional but valuable), second major paper
Year 5	Dissertation writing, job search, final defense preparation

See checklists/phd-milestones-checklist.md for a detailed, actionable milestone tracker.