"Up and coming." "Promising." "One to watch." Every job board and every LinkedIn feed is full of people describing themselves with adjectives. I do not want to do that, because adjectives are unverifiable noise. So this post is something else: my first year of building in public, in receipts, with links. Judge the trajectory, not the adjectives.
The receipts
2022 — the degree, done at night. I started a BSc in Computing and IT with the Open University while working full time. I finished it in 2026. Not a receipt for talent — a receipt for stubbornness, which is more reliable anyway.
2025 — n8n templates that run in production. I built a set of workflow templates for n8n and put them to work. The alert monitor has run 1,892 times at 99.7% success. Lead qualification: 247 runs, 99.2%. Data sync: 312 runs, 97.8%. They are open source: n8n-automation-templates. A template that runs weekly for months without you touching it is worth more than any certificate.
2026 — the job search engine. I automated my own job hunt: 30+ ATS boards, deduplication against every posting I have ever seen, hard rule filters before any LLM call, then LLM scoring against my honest profile. Over 1,500 postings evaluated this year; 1,539 parked; 12 apply-worthy. Public and scrubbed: remote-job-engine and job-sweep. The full story is on dev.to, and auf Deutsch too.
September 2026 — the quality wave. One month, four repos, all with tests: a composable ETL framework (errors as data, whitelist mapping, CI), a SQL playground (55 templates, 115 tests, now a live demo), a file automation toolkit (20 tests), and a finance analytics dashboard. The lesson post is here.
October 2026 — the paper trail. Three technical articles, a bilingual portfolio site, and a live demo instead of dead links. You are reading the third article.
What the receipts have in common
Build things you actually use. My n8n templates monitor services I care about. My job engine found the shortlist I am actually working through. Tools you use daily cannot fake it.
Tests or it did not happen. Every framework repo has a test suite. The SQL playground has 115 tests that execute all 55 templates. This is not rigour for its own sake — it is the difference between "code exists" and "code works."
Write it down. The articles are not marketing. They are the honest lessons, including the ones that cost me: six perfect-looking job candidates that died when I read the actual posting bodies.
The honest gap
I have no professional software engineering title yet. Thirteen years in operations — Lloyds Banking Group, Deutsche Post, sysadmin, currently Amazon print production — and a degree, and all the receipts above, and no job title in software. That is the next receipt, and it is the one I cannot publish myself. Someone has to give it to me.
If your team needs a junior or associate who already works like this — AI, automation, or data, fully remote from Germany or UK — everything is on reaver1000.github.io. The receipts are the résumé. The next one is yours to write.
Top comments (1)
I like the “receipts instead of adjectives” framing, and the honest gap section is the strongest part. Tests plus things you actually use every day is a convincing combination.
A few suggestions that might make the receipts land even harder with hiring managers:
Add an evaluation story to the job engine. You score postings with an LLM, so the natural next receipt is measurement. For example, label a few hundred postings yourself and report precision and recall for the “apply-worthy” class. That’s the skill AI teams hire for, and your “six perfect candidates died” lesson is already half of it.
Lead with one flagship, not five repos. Pick the project closest to the role you want, and give it a one-paragraph story: problem, approach, result, and what you’d change. The rest can sit behind it as supporting evidence.
Use your operations background as an advantage. Thirteen years in banking, logistics, and production at that scale is rare among juniors. Roles like automation engineer, data engineer, analytics engineer, and technical operations value people who know how real processes break.
Show the failure side of your n8n numbers. You report success rates of 97.8–99.7%, and a short note on what the failures were and how you handle them (retries, alerting, idempotency) signals production thinking.
Good luck with the search. Which kind of role are you targeting first: automation, data, or AI?