Context: Full Sail B.S. student + AWS intern documenting how I study. These tactics come from my Notion tracker, not an official training team.
AI assist: ChatGPT helped outline the sections and summarize Study Hall notes; I reconciled everything with my actual tracker on 2025-10-15.
Status: Early-career roadmap. Certifications prove I can learn, but they don’t replace production experience—so I pair every badge with a project.
Reality snapshot
- Active certs: AWS Solutions Architect – Associate (Aug 2024–Aug 2027) and AWS Certified AI Practitioner (Feb 2025–Feb 2028).
- Supporting coursework: freeCodeCamp JS Algorithms + Responsive Web Design, plus LinkedIn Learning soft skills (communication, GTD, personal branding).
- Tracker: Notion database with columns for domain, study hours, lab status, exam date, renewal reminders, and “paired project” (the thing I’ll build to prove it).
My study pipeline
1. Start with the exam guide
- Copy the official outline into Notion. I color-code each domain: green (confident), yellow (needs lab), red (needs deep dive).
- Add sample question links + whitepapers per domain. If I can’t explain the whitepaper to a friend, it stays red.
2. Plan the reps
| Practice type | Tools | Notes |
|---|
| Hands-on labs | AWS Skill Builder, Sandbox accounts | Spin up services, break them on purpose, then fix using CloudWatch + docs. |
| Flashcards | Anki decks per domain | Focus on limits, IAM condition keys, service integrations. Spaced repetition handles the rest. |
| Teach-backs | Loom recordings, study group | If I can’t explain a topic in 5 minutes, I haven’t learned it. I re-record after each round. |
| Mock exams | Tutorials Dojo, AWS practice tests | Take them cold once per week. Anything below 80% becomes next week’s lab. |
3. Pair a project with each badge
- SAA-C03: Built a Well-Architected checklist into Car-Match + the AWS internship capstone. Every section (Ops, Security, Reliability, Performance, Cost) gets a bullet in the repo README.
- AI Practitioner: Logged every Bedrock experiment (prompt templates, guardrails, cost estimates) in
notes/bedrock/. The goal wasn’t to become an ML engineer—just to show I can wire AI responsibly.
- freeCodeCamp certificates: Used the JS Algorithms work to refresh LeetCode patterns and the Responsive Web Design course to audit this portfolio’s layout.
4. Close the loop
- Retro doc: After each exam, I write a short postmortem: what worked, what didn’t, what to revisit in 3/6/12 months. Lives in
notes/certs/<exam>.md.
- Renewal reminders: Notion triggers at T-6 months, T-3 months, and T-1 month. Each reminder includes the “paired project” I’ll refresh (e.g., re-run cost optimization labs, rebuild a CI/CD pipeline).
- Public accountability: One LinkedIn post per milestone. Not for clout—just to keep friends/mentors in the loop.
Current roadmap
| Milestone | Target | Supporting project |
|---|
| AWS Developer Associate | Feb 2026 | Rebuild CheeseMath backend with AWS SAM + full test coverage. |
| AWS Security Specialty (stretch) | Late 2026 | Document IAM guardrails + incident drills for personal projects. |
| Zig + WebGPU deep dives | Ongoing | Expand Triangle Shader Lab and OBJ Parser with performance benchmarks + wasm builds. |
| Soft skills refresh | Quarterly | Run peer workshops on documentation, async updates, and honesty audits. |
Evidence & templates
Lessons learned
- The exam isn’t the finish line; the project you ship afterward is.
- Honest notes matter more than badges. Tracking energy, sleep, and morale kept me from cramming myself into burnout.
- Share the playbook. Letting classmates copy my Notion setup keeps us all accountable and surfaces gaps I missed.
Daily cadence that actually sticks
- Morning (45–60 min): One domain review + 15 flashcards + a quick “teach-back” recording to force clarity.
- Lunch (20 min): One practice question set, then tag anything below 80% as “lab tonight.”
- Evening (60–90 min): Lab or mock exam. I end the day by updating the tracker with mood, sleep, and whether my guesses were lucky or backed by understanding.
- Weekly retro (30 min, Sundays): Rename every “maybe later” task to either “schedule” or “delete.” If I can’t defend why it matters, it goes.
- Rest gates: No study if sleep < 7h, if caffeine > 400mg, or if morale log says “fried.” Those nights are for walks, not whitepapers.
Common traps I hit (and how I patched them)
- Shiny course hopping: I kept buying new courses when stuck. Now I set a “one-course-per-domain” rule and only switch after a written retro.
- Passive video watching: Watching lectures felt productive but stuck nothing. The fix: no video without a parallel lab or notes doc open.
- Ignoring quotas and limits: Mock exams punished me on nitpicky service limits. Solution: an Anki deck labeled “limits-only” drilled daily.
- Cramming before renewals: Renewals snuck up on me until I built the T-6/T-3/T-1 reminders and paired projects.
- Letting AI hallucinate details: I now annotate AI-summarized whitepapers with page references; if I can’t find the claim in the source, it gets deleted.
Budget and time guardrails
- Money: $0 if I stick to Skill Builder, free tiers, and library books. ~$80/month when I add Tutorials Dojo + a paid mock set. I log every purchase in Notion so I know the burn rate.
- Time: Max 10 study hours per week during school terms, 15 during breaks. Any more and grades or sleep suffer; the tracker auto-flags if I exceed.
- Energy: I rate each session green/yellow/red. Two reds in a row means a break day; studying while exhausted just cements confusion.
Proof I can apply it (interviews and projects)
- Mock interview prompts: I practice “Explain S3 durability vs availability” and “Design a cheap static site with auth” aloud on Loom, then rewatch to trim fluff.
- Portfolio receipts: Each project README lists the cert domain it covers (e.g., Car-Match → VPC, IAM, ALB basics). Recruiters see exam knowledge tied to running code.
- Numbers: I log mock scores by domain (e.g., SAA Week 3: Security 68%, Reliability 72%). Improvement charts give me concrete proof I’m progressing, not just “feeling” ready.
- Behavioral stories: I keep STAR notes for “When I broke prod in a lab and fixed it,” “When AI was wrong and I caught it,” and “When I said no to scope creep during study.” They map to interview questions quickly.
How AI fits without replacing the work
- Outline aid: Ask ChatGPT for a first-pass outline, then prune aggressively to match the official guide.
- Flashcard seed: Generate sample questions, but every card gets reviewed against AWS docs before entering Anki.
- Lab buddy: Use Copilot for YAML scaffolds (SAM/CloudFormation), then lint and diff against docs.
- Guardrails: Any AI-generated claim must include the doc link; if not, it’s suspect. I keep a
hallucinations.md with the worst offenders as reminders.
What changes after each badge
- Week after passing: Write the retro, publish a short LinkedIn post, and schedule the paired project. No new certs until the project ships.
- Month after: Re-read weakest domain notes and ship one small improvement (e.g., add alarms, add budgets, improve tagging).
- Six months after: Re-run a mock exam cold. If the score dips below 80%, schedule a refresher sprint; otherwise, resume normal study load.
Open questions I’m still working on
- Best way to balance deep Linux/OS study with cloud time without burning both ends.
- How to keep AI summaries lean without losing nuance from long whitepapers.
- Whether to tackle Security Specialty before or after a heavier dev-focused cert.
- Finding a repeatable way to publish lab writeups without slowing school deadlines.
References