arm
  • Home
  • Work
  • About
  • Writing
  • Resume
  • Contact
Open to work
❡Case study · Deep dive · 8 min read

NxtRmtJobs — a remote board that knows about your timezone.

For Indian developers chasing global remote roles: the exact IST overlap window, computed per listing.

buildingpersonalpersonaljobsbuilding
The overlap window, front and centre
Every remote job board shows salary and stack. None of them answer the question an Indian developer actually asks first: how many hours of my day overlap with this team? NxtRmtJobs computes the IST overlap for each role and puts it where the salary usually goes.
IST
Anchor timezone
FTS
Search engine
1
Own backend
24/7
Ingestion
Listing with overlap meter

§01The problem

A “remote” role advertised from San Francisco can mean a 9pm–2am working day from Ludhiana. That single fact decides whether a job is livable, yet it is never on the listing — applicants reverse-engineer it from a city name after they have already invested in applying.

NxtRmtJobs treats timezone overlap as a first-class filter, the way other boards treat salary band or seniority.

§02System architecture

A real backend, not just a Next.js API folder — Hono owns ingestion, search, and the overlap computation, with Postgres full-text search doing the heavy lifting.

Client
Next.js 15Overlap meter UITailwind v4
API
HonoOverlap calculatorBetter Auth
Search
Postgres full-textDrizzleRanked results
Ingestion
BullMQ jobsScheduled scrapesDedup + normalise

§03Three decisions worth defending

Decision 01

Overlap as a computed, indexed field

The IST overlap is derived once at ingestion from the role's stated timezone and stored, so it is sortable and filterable instantly — not recomputed in the browser on every render.

Decision 02

Postgres FTS over a search service

Bringing in Elastic/Meilisearch for a single-developer project is premature infrastructure. Postgres full-text search with proper indexes handles the corpus comfortably and keeps the stack to one datastore.

Decision 03

Own Hono backend over Next API routes

Ingestion, queues, and search wanted a long-running service with its own lifecycle. A dedicated Hono backend keeps that work out of the serverless request path and gives the typed hc client to the front end.

§04Hardest problems, plainly

Challenge 01

Timezones are a swamp

DST, half-hour offsets, and listings that state a city instead of a zone all conspire against a clean overlap number. Normalising messy source data into a trustworthy IST window is the core difficulty.

Challenge 02

Keeping ingestion clean

The same role appears on multiple sources with slightly different titles. Deduplication and normalisation at ingestion time decide whether search results feel curated or like noise.

Challenge 03

Ranking that respects overlap

Relevance is not just text match — a perfect keyword hit at 3am IST is worse than a good hit at noon. Blending FTS rank with overlap into a single sensible ordering took iteration.

“Salary tells you if a remote job is worth it. Timezone tells you if it is survivable. I built the board around the second number.”— Parm, on NxtRmtJobs
Talk about this project →
Open to work

Have something in mind? Let's make it.

iamparmjeetmishra@gmail.com→
GithubMailLinkedInX/Twitter
arm© 2026 — Made with ♡ by Parm
HomeWorkAboutWritingResumeContact