Hard skills

The craft layer of Career development — the technical abilities that produce the work in the first place. Unlike Soft skills (leverage), hard skills compound through depth: the spike model from swyx’s Coding Career Handbook — differentiated depth in one area beats even spread across many.

The garden is already full of hard-skills material; this note is the shelf that links it:

• Systems & infrastructure: Homelabs, Self-hosted CI-CD, Golden repo (the reference-repo rules), Diff tools, Git hooks
• Local AI / inference: Local LLM inference, Inference engines, Qwen model family, Fine-tuning GLiNER
• Testing discipline: Agent tests, LLM evals, Property-based testing
• Data engineering: Change Data Capture, DuckDB-side captures, Zod vs Valibot-style schema literacy
• Security: PII redaction, Agent sandbox, Uncensored models

Growing: (open — which spikes to invest in next quarter; the T-shaped-vs-pi-shaped question; certification ROI in COT context)

Related: Career development · Soft skills