Badges are derived deterministically from the evaluation — earning rules are public code, no manual award. Each tile links to the evidence that produced it.

§ 01 · Idea → MVP

What I can ship

Simple MVP (auth + DB + 3-5 features) in 2-3 days

Reference project
aihost.md — from initial commit to production with multi-tenant + auth + billing in ~7 days (Apr 2026)

With AI2-3 days

Without AI1-2 weeks

Confidencehigh

§ 01b · Hire-fit signals

Where I fit

Role type

Founder

Team setting

Solo

Build mix (last 90d)

70% greenfield · 30% maintain

Domain verticals

B2B SaaS · dev tools · marketplace

Stack depth

Stack	Months	Repos
Next.js	18	6
Drizzle	8	4
Postgres	14	5
Better-Auth	4	3
Tailwind	18	6

§ 01c · Quality discipline

How I keep code from rotting

Tests

No tests yet

Refactor cadence

12 refactors / month

Scope tends to: extraction

Docs as output

README 75/100

11 ADRs · 22% inline doc ratio

Agent safety

Allowlist: 334 entries

Secrets handling: good · scoped agents

Bug-fix vs feature

32% fixes

Balanced — active iteration

§ 02 · Deliverables

What I can build for your business

What	Category	With AI	Without AI	Confidence
Multi-tenant web app with authentication + database aihost.md in production: Next.js 15 + Better-Auth + Drizzle + per-user Docker — repeatable pattern	Web	2-3 days	1-2 weeks	high
Campaign page with form + admin panel /scrisoare deployed with form actions + Tiptap editor + admin — reusable pattern	Marketing	1 day	1 week	high
Technical documentation + sprint execution plan Detailed CLAUDE.md (134 lines), instructions/ with 11 numbered files, RESUME.md, active plans/ folder	Docs & planning	a few hours	2-3 days	high
Minimal deployment setup (Docker + reverse proxy) Structured deploy.sh, infra/ folder with compose, container orchestration aihost-postgres + live services	DevOps	1 day	3-5 days	medium
Refactoring + framework modernization Git history shows systematic polish/refactor commits; inline comments for learnings from real incidents	Other	3-5 days	2-3 weeks	medium

§ 03 · Achievements

What I shipped, with timeline

7 days

aihost.md in production — multi-tenant + auth + billing

Next.js 15DrizzleBetter-AuthDocker

deployed live, paying users, multiple servers

2 days

/scrisoare campaign page — campaign + admin + Tiptap editor

Next.jsserver actionsTiptap

reusable pattern for other campaigns

3 days

Top AI People — leaderboard with versioned prompt

Next.jsPostgresDrizzleHMAC tokens

5 prompt versions iterated with user feedback

§ 04 · Measured outcomes

Numbers, not vibes

7 days

aihost.md from initial commit to production

own SaaS

5 prompt iterations

v1.0 → v1.5 with user feedback baked in

top-ai-people feature in 3 days

334 entries

permission allowlist .claude/settings.local.json

discipline built over time

Fastest end-to-end delivery

/scrisoare page — form + admin + Tiptap, deployed · 2 days

Signature bug fix

Drift between Drizzle journal and DB schema after drizzle-kit push instead of migrate — manually backfilled SHA256 hashes to restore tracking

With AI: 45 min · Without AI: half a day

§ 05 · Collaboration style

How I work with AI

Fast iteration with clear commits — git log shows tight AI→commit cycles with structured messages (feat/fix/polish/docs)
Mature context engineering — CLAUDE.md contains specific rules derived from incidents (fail2ban ban, DB drift)
Permission discipline — 334-entry allowlist built over time, no permissive auto-approve
Uses planning workflow — superpowers + active plans/ folder for complex projects

§ 06 · Ship throughput

Finishing rate

83%

completion rate

started: 6

shipped to MVP: 5

abandoned: 1

period: 6 months

Memory + git: aihost.md · /scrisoare · top-ai-people · cardots-hetzner deploy · revoluția-silențioasă · plus 1 abandoned. 5 deployed with public URL, 1 internal.

§ 07 · Toolchain

Tools I reach for

FigmaLinearGitHubDockerPostgresMailHogBullMQCursorTiptapsuperpowers (Claude plugin)

§ 08 · AI Fluency, broken down

Score components

Output quality78 / 100

Ship rate78 / 100

Context engineering68 / 100

Iteration patterns75 / 100

Solid ship rate: aihost.md in production · /scrisoare deployed · top-ai-people live with paying users
Output quality: Next.js 15 + Drizzle + Better-Auth integrated cleanly, disciplined schemas
Context engineering: per-project CLAUDE.md (134 lines with rules derived from real incidents: fail2ban, DB drift)
Iteration patterns: git shows clear AI→commit cycle, structured messages (feat/fix/polish/docs/refactor)
Tool-agnostic: pattern reproducible with Cursor/Codex — focus on output, not Claude-specific

§ 09 · Skills

Tagged proficiencies

TypeScriptadvancedNext.jsadvancedDrizzle ORMadvancedTailwind CSSadvancedPostgreSQLintermediateBetter-AuthintermediateDockerintermediateAI workflow designadvanced

§ 10 · Verdict

What stands out

01Mature context engineering — CLAUDE.md with rules derived from real incidents (fail2ban, DB drift, deploy safety)
02Disciplined AI setup — 334-entry permission allowlist built over time, no permissive auto-approve
03Production discipline — Postgres in production, backups configured, structured deploy.sh
04Clean stack engineering — Next.js 15 + Drizzle + Better-Auth integrated correctly, well-structured schemas