Resources

Sample outreach

Three worked examples of what the Outreach Writer drafts — across cold, follow-up, and thank-you modes. Every draft is yours to edit before sending.

ColdFirst touch, no prior contact.

Scenario — Undergrad applying to an NLP interpretability group; references the professor's NeurIPS 2025 paper on probe calibration.

Subject
Probe calibration + your NeurIPS 2025 paper — F26 applicant
Body
Dear Dr. Chen,

I'm a final-year undergraduate at <your school> working on attention-based interpretability. Your recent paper on probe calibration in open-ended generation resonated with a line of work I started at a 2023 Meta internship — specifically, the angle on causal vs correlational probes.

I'd love to apply to your lab this F26 cycle. I've attached my CV. If there's time in the next two weeks, I'd be grateful for a 20-minute conversation about where the probe-calibration agenda is headed.

Best,
<your name>

Why it works

  • •Names the specific paper and the angle — signals you actually read it
  • •Asks for 20 min, not an hour — realistic commitment
  • •Closes with a concrete time window, not an open-ended ask
Follow-upPolite nudge after two weeks of silence.

Scenario — Gentle follow-up on a cold email that got no reply.

Subject
Following up — probe calibration outreach
Body
Dear Dr. Chen,

Following up on the email below — I understand this is a busy stretch.
If a 20-minute call won't land this cycle, I'd still be glad to share a short memo on where I think the probe-calibration work could extend.

Happy to hold off if the timing is wrong.

Best,
<your name>

Why it works

  • •Acknowledges the recipient is busy — no guilt-tripping
  • •Offers an asynchronous alternative (a short memo) that costs the recipient zero time
  • •Gives a polite exit — "happy to hold off"
Thank youSent within 24 hours of a meeting.

Scenario — After a Zoom call where the professor flagged an open project on calibration under distribution shift.

Subject
Thank you — and a quick follow-up on Tuesday
Body
Dear Dr. Chen,

Thank you for the time on Tuesday. The open thread on calibration under distribution shift is exactly the kind of problem I'd want to spend three years on — I went back and re-read Section 4 of the NeurIPS paper with that in mind.

A quick note on what we discussed:
— I'll draft a short write-up on the OOD-calibration angle and send it by end of next week.
— My CV is attached in the thread below for easy reference.

Thanks again — I'll be in touch.

<your name>

Why it works

  • •Specific recall of what was discussed — shows the meeting registered
  • •Names a concrete next step with a date
  • •Stays short — respects the recipient's inbox