For many people, the difficult part is not knowing what happens after sending the message. You can begin by asking a specific question, learning about a project, or exploring a possible collaboration.
If you are drafting your first email, use the example below and replace the topic, prior attempts, and request with your own.
Find a reason to talk
Read the person's recent papers, projects, or open calls. Know what interests you, what you have tried, and what you want to ask.
Stanford's guide to emailing faculty offers a useful model. Local instructors, students at other universities, and paper authors can all be people to talk to.
The first email can be short
Adapt the details in this example to your own situation:
Hello Professor, I am an undergraduate studying [subject] at [university], currently exploring long-horizon agents. After reading [paper] from your group, I became interested in its failure-recovery component. I tried the related example and wrote a one-page record.
May I ask whether you have a task suitable for an undergraduate, or time for a short conversation about this question? I can currently commit [time] each week. My notes and CV are attached. Thank you.
State a specific request, such as checking an experimental setting or suggesting related work, so the recipient can assess the topic and time involved. If there is no reply, follow up once after a reasonable interval and continue checking other relevant contacts.
What to clarify in the first conversation
Ask about the actual task, supervision, communication, data, compute, and time. Explain your aims and foundations so both sides can assess the fit.
For a trial, agree on when to review its outcome. Discuss contribution, authorship, and handover expectations early and update them as the work changes.
How to ask when stuck
Expand “the program will not run” into what you wanted to execute, the environment, where it failed, and what you tried. Research questions similarly need the goal, observation, attempts, and hypothesis.
The information-organization sections of How To Ask Questions The Smart Way are useful. How to work with a professor also discusses records, updates, and collaboration.
Adjustments and handoff
If scope, time, or resources differ from the agreement, list the differences and discuss adjustments with the project contact. When ending the collaboration, confirm the end date, finished and unfinished tasks, code and data locations, handoff recipient, and use of outputs.
See Q25 for handoff checks and questions 16–25 for contact details.
Other accounts of contacting and collaborating
Research conversations includes Jason Eisner's advice, and Stanford's email guide. If you do not yet know which groups exist, follow projects through the lab directory.