client portal prometrifi
← all projects
case study · career automation

Career Feed Bot

Finds the roles worth applying to, scores the fit, and has a tailored resume ready before you've finished reading the posting.

Renaissance-style illustration of two figures examining a wall of sealed notices and postings, representing the daily job-board scan, in Prometrifi's dithered Digital Renaissance style.
the gap

Checking six job boards by hand is a part-time job on top of the one you're trying to land.

A real search means going source by source every morning, skimming past everything off-target, and writing a fresh angle for each listing that actually fits — before someone else applies first. Most of what's posted isn't worth a look. The handful that are get buried in the same routine as everything else.

how it runs

One scan, one score, one click to apply.

01

Scrape

Every morning at 8am, six sources — HYVE, Freelancer.com, CryptoCurrencyJobs, LaborX, HireWeb3, and the Web3 Career API — are pulled in parallel and normalized to one shape.

02

Score

New listings are deduped against what's already been seen, filtered for obvious misses, and scored against a target profile by a locally-run model — every score comes with a stated reason.

03

Digest

Anything scoring 65% or higher, with a link that's still live, lands in a Slack digest with its match score, the reasoning, and the original posting attached.

04

Apply

Accepting a listing in Slack drafts a tailored resume on the spot, renders it to a real PDF back in the same thread, with a matching cover letter one click behind it.

built with

The actual stack.

n8n workflow automation Ollama (Hermes) local scoring Claude API resume/cover letter drafting Gotenberg HTML→PDF rendering Postgres seen/applied tracking Slack interactive digest
where it stands
Runs daily at 8am across six sources — live-link checks and dedupe keep the digest to only what's new and actually open.
the architecture

One pipeline, 4 stages.

01 Scrape Six sources pulled daily, normalized to one shape.
→
02 Score Deduped, filtered, and scored with a stated reason.
→
03 Digest 65%+ matches with a live link land in Slack.
→
04 Apply Tailored resume and cover letter, one click away.
proof of work — the pipeline

The n8n canvas, in full.

n8n workflow canvas: a daily 8am trigger scraping HYVE, Freelancer.com, CryptoCurrencyJobs, LaborX, HireWeb3, and the Web3 Career API, merging and aggregating listings, deduping against Postgres, filtering obvious misses, scoring with Ollama (Hermes), and routing anything at or above a 65% match to a Slack digest.
Every morning at 8am: six sources scraped in parallel, merged, deduped against what's already been seen, scored by a local Ollama model, and anything at or above a 65% match — with a live link — posted to the digest.
proof of work — sample output

What lands in Slack.

Slack messages from Career Feed Bot showing job listings with a match percentage, a stated reason for the score, an unfurled posting card, and Accept & Tailor Resume / View Posting buttons.
Each listing arrives with its match score and the reasoning behind it, the original posting attached, and one button to accept and start tailoring — no separate tab required.
proof of work — the apply flow

From click to a real PDF.

n8n workflow canvas: a Slack interactivity webhook parsing an Accept or Cover Letter action, pulling job details, drafting a tailored resume with Claude, rendering it to PDF via Gotenberg, uploading it back to Slack, and offering a cover letter button.
Accepting a listing in Slack fires this flow: pull the job details, draft a tailored resume with Claude, render it to a real PDF, and upload it straight back into the thread — with a cover letter one button behind it.
let's work together

Have a project in mind?

hello@prometrifi.co
← back to all projects