← Back to Portfolio
Journey: QA Lead → Prompt Engineer in 4 Weeks
Value Prop: "I turn vague requirements into reliable LLM features—measured, cost-optimal, shipped."
Goal
Get interview-ready for a Prompt Engineer role in 4 weeks through a structured, high-impact learning plan.
Plan Overview
This journey focuses on what hiring managers actually test:
- Prompt craft - Writing effective prompts
- Eval & metrics - Proving prompts work with data
- Domain transfer - Adapting to new domains
- Tooling/automation - Building production-ready systems
- Communication - Articulating decisions with data
Time Commitment: ~10 hours/week (90-minute daily blocks)
Week-by-Week Breakdown
Week 1: Core Prompting
Build reflex-level prompting hygiene
View Plan →
Week 2: Evaluation & Data
Prove prompts work with metrics
View Plan →
Week 3: Domain Transfer
Ship production-ready prompts
View Plan →
Week 4: Portfolio & Interviews
Package evidence for recruiters
View Plan →
Checkpoint Targets
- Public GitHub repo ≥50 commits, ≥3 runnable projects
- At least one write-up with >1k views or 50 claps
- Can recite exact BLEU score improvement + cost delta for every project without notes
- Five AI-scored answers ≥8/10 + one 45-min peer session with written feedback
Resources
Datasets
- GSM8K (math reasoning)
- MT-bench (chat evaluation)
- PubMedQA (medical Q&A)
- CaseHOLD (legal reasoning)
- SQuAD (reading comprehension)
Tools
- Promptfoo - Prompt testing framework
- Phoenix Arize - LLM observability
- LangSmith - LangChain debugging
- Weights & Biases "Traces" - Experiment tracking
Ready-Mode Checklist
- Portfolio live
- 3 case studies with numbers
- GitHub pinned repos
- Resume bullet: "Improved LLM task X from A% → B% accuracy while cutting cost by C% via prompt optimization"
- Calendly link in email signature for recruiter calls
View Full Project on GitHub →