research-perks-radar 🛰️

An open, auto-updating radar for research perks — free AI/compute credits, funding & grants, academic software discounts, datasets, awards, and travel support. Scanned daily, scored automatically, and matched to your CV right in your browser.

Live site: https://jiaye1998.github.io/research-perks-radar/

119 live perks · ai_compute: 27 · funding: 26 · software: 4 · data: 26 · awards: 16 · events: 20 · updated 2026-07-19

Use with Codex

This repository includes a Codex Skill for working with Research Perks Radar.

In Codex, ask:

Install the Codex Skill from https://github.com/Jiaye1998/research-perks-radar/tree/main/skills/research-perks-radar

Then restart Codex or open a new thread so the Skill is picked up.

After installation, you can ask Codex things like:

Use the Research Perks Radar Skill to refresh the data and run the checks.

Use the Research Perks Radar Skill to update the web output after a data refresh.

---

What it does

Every day, a GitHub Action runs a zero-LLM Python pipeline that:

1. Fetches candidate perks from four kinds of sources — curated sites, free search APIs (Tavily + Brave), RSS feeds, and Reddit.
2. Processes them with a pure rule engine — dedup, classify into 6 categories, extract amount/deadline/region, filter noise, score, and flag deadlines.
3. Publishes data/perks.json and rebuilds a clean, Apple-style Next.js site deployed to GitHub Pages.

CV matching happens entirely in your browser. You paste your CV and your own LLM API key (Anthropic / OpenAI / Google); the page calls the model directly and re-ranks perks by fit. Your CV and key never leave the page and never touch our servers — there are no servers.

The six categories

Category	What's in it
ai_compute	Free AI tool credits, API quotas, HPC/cloud compute for researchers
funding	Grants, fellowships, seed funding, travel awards
software	Academic licenses & discounts (GitHub, JetBrains, Overleaf, …)
data	Dataset access, infrastructure programs
awards	Scholarships, prizes, postdoc fellowships
events	Conference fee waivers, travel grants, summer schools

Repo layout

.github/workflows/daily.yml   # cron + build + deploy
pipeline/                     # Python backend (no LLM)
  main.py                     # orchestrator
  sources.yaml                # data sources — PRs welcome!
  fetchers/                   # curated / search / rss / reddit
  rules/                      # classify · extract · filter · score · urgency
  dedup.py
data/perks.json               # main feed the site reads
data/history/                 # daily snapshots
web/                          # Next.js static site

Run the pipeline locally

cd pipeline
pip install -r ../requirements.txt
# optional free keys for richer results:
export TAVILY_API_KEY=...      # tavily.com  (1000/mo free)
export BRAVE_API_KEY=...       # brave.com/search/api (2000/mo free)
python main.py                 # writes ../data/perks.json

No keys? It still runs — curated sources, RSS, and Reddit need none.

Run the site locally

cd web
npm install
npm run dev      # http://localhost:3000

Contributing sources

Add an entry to pipeline/sources.yaml and open a PR. See the comments in that file for the schema.

License

MIT