Add New Faculty Member
Color guide
Pipeline — how each monthly digest is built
Two ways to read the highlights
The same data, pivoted. Switch with the toggle at the top left of the Faculty Highlights view.
- 📅 By month — every faculty member for one month. Answers “what happened in August?” Pick the month from the tabs; older months are in the … older dropdown.
- 👤 By faculty — one faculty member across several months, still grouped month by month. Answers “what has this person done lately?”, which is the question when writing a nomination, a bio, or an introduction. The Window control covers the last 3, 6 or 12 months, or all of them.
In By faculty mode the search box lists every faculty member, ordered by how many items they have in the chosen window, so the people with something to show are at the top. The number beside each name is that count, and a divider marks where the people with no items in the window begin. Type to filter by name, department or title.
Months with nothing are still shown, marked “nothing recorded”. That is deliberate: a silent gap looks like missing data, whereas a stated gap tells you the pipeline ran and found nothing. A month that fails to load says so and is left out of the totals.
Shortcut: each card in Faculty Profiles has a 📰 Highlights link that jumps straight to that person's view. The ⬇ Word / Excel / CSV exports always match what is on screen, so exporting in By-faculty mode gives you that person across the window.
Models in use
Monthly digest — verification (S3) and summarization (S4). Fast and cheap; sufficient for structured fact-checking and short summaries.
Quarterly highlights — curation and narrative generation. More capable model used where output quality matters most.
All calls go through the Anthropic API. No fine-tuning; standard prompt engineering only.
Model versions are pinned in shared/digest_guardrails.py.
Data sources
- arXiv — preprints (cs, eess, q-bio, stat)
- Google Scholar via SerpAPI — papers, citations, news mentions
- PubMed — biomedical publications
- EurekAlert — university press releases
- Faculty websites — scraped per-person (author_specific)
- LinkedIn via Lix API — posts, announcements (credit-limited)
SerpAPI and Lix credits are consumed per run. LinkedIn enrichment may be skipped
to conserve credits (--skip-linkedin flag).
Item provenance badges
Each item in the digest carries a badge showing where it was found:
author_specific). Highest-trust source.
scraped_webpage).
general). Requires stronger LLM verification.
Hover any item's badges to see the LLM's verify_reason.
Known limitations
Pipeline — how each faculty profile is built and maintained
Models in use
All profile pipeline steps — web enrichment, theme extraction, keyword extraction, bio rewriting. Fast and cost-effective for structured extraction tasks.
Faculty profiles run on a weekly schedule and on-demand via "↺ Update Profile." Incremental mode skips LLM steps if the source data hasn't changed.
Known limitations
Pipeline — how a synergies match is computed
Models in use
Material compression when inputs are large. Fast, cheap, used only when needed.
Entity extraction and executive summary — the core matching intelligence. Highest-capability model; costs reflect this (~$0.05–$2 per project depending on materials size).
Cost and CO₂ estimates are shown on each project card after a run.
Known limitations
Pipeline — how the weekly scan finds a call
Models in use
Reading pages, screening calls, filling in detail. This is the bulk of the work, a few hundred small calls per run.
Matching calls to faculty, where the judgement is worth paying for.
A run costs roughly a dollar. The figure for the most recent scan is shown on the Funding tab itself.
What it decides, and what it does not
It decides which calls are worth your attention and which faculty to name against each one. Both are judgements, and both are shown with the reasoning attached so you can disagree with them.
It does not file anything, contact any funder, or commit you to a submission. Every deadline, amount and eligibility line is read off the funder's own page, and that page is one click away on every card. Check it before you rely on it.
A call whose deadline cannot be parsed is kept, not dropped. Not knowing that something closed is not evidence that it did, and a missed call is the more costly mistake.
Where the two lists come from
New Funding Calls is what this week's run found. Open Funding Calls is everything found in earlier weeks whose deadline has not yet passed, because a strong match from three weeks ago is no less worth applying for, it has simply stopped being news. A call appears in one list or the other, never both.
The scan report lists every source it read, including the ones that returned nothing, so a funder going quiet is visible rather than indistinguishable from a funder whose page broke.