AgileTech Solutions · Amsterdam
Query the engineer.
I run AgileTech Solutions, where I build the parts nobody screenshots: schemas, resolvers, indexes, and the operational work that keeps them up. This page is served by the same kind of API I'd build for you — so read it, or just ask it.
Node 01 · explain analyze
A feed query that got 28× faster, and why.
The marketplace listing feed joined sellers and their active offers. It worked at 200 rows and fell over at 40,000. Here is the plan before and after.
Diagnosis
Prisma issued one offers query per listing. The join column had no index, so each of those was a sequential scan over the whole offers table.
Change
A single batched fetch replaced the per-row loop, plus a composite index on (listing_id, status) matching the filter and the sort.
Cost
The index adds write overhead on a hot table and roughly 40 MB. Reads outnumber writes here about 90 to 1, so it pays for itself.
Node 02 · systems
Four systems, and what each one cost me.
Every decision bought something and charged something. Both columns are filled in — pick a system to read the entry.
Node 03 · mutation
Send me something to build.
This form posts to an API I wrote and run myself — the same service handling submissions for two client sites in production. No form service, no embed.
Required.
Needs a valid address.
A sentence or two, at least.