The Deeply Impersonal Personalized Recruiter Mail – Pentlander Blog



The Deeply Impersonal Personalized Recruiter Mail – Pentlander Blog

















Lately I’ve encountered a brand new species of software program engineer recruiter mail. It goes one thing like this:

Your work on development projects at Railway, alongside together with your earlier years constructing observability software program at City Storage Systems and storage techniques on AWS Glacier, is precisely the sort of platform background this function requires.

So the primary paragraph is spent explaining my profession to me (thanks I forgot the place I labored). The final sentence is coincidentally “that’s precisely what we’re searching for”! Then follows the precise job description for constructing CI/CD pipelines, which has nearly nothing to do with the job expertise simply talked about. Clearly the primary paragraph is generated by an LLM, however the job description too?!

A couple of gems I’ve gotten:

That’s the hole. You’d be the primary. The work is concrete: multi-tenant isolation for AI mannequin serving, golden-path service templates in Terraform, IAM and secrets and techniques administration wired into CI/CD, and policy-as-code that makes safe the default relatively than the exception.

Observability right here is genuinely unsolved. The failure modes are probabilistic relatively than deterministic, so customary instrumentation tells you lower than you want.

Every doc is a referral, a previous auth, or an consumption type, which implies extraction accuracy is not a dashboard metric, it is whether or not a affected person will get seen or will get denied. Messy inputs, no tolerance for silent failure, rising quick.

Over the years I’ve gotten plenty of emails alongside the traces of “Hey it’s actually neat that you just work at X firm” instantly adopted by the job description, so what’s the distinction? So what if it’s generated by an LLM?

To me the distinction is twofold:

  1. “Personalized” with zero connection to the job description
  2. Being very apparent LLM textual content

With the previous, I perceive recruiters ship a number of emails to hit the numbers they want. The e mail template with [insert name and previous company] is a shotgun blast, they in all probability have not spent greater than 10 seconds taking a look at my profile to see if it’s a match. So it’s unsurprising to me that it’s for a Django fullstack function when my historical past exhibits I’m an infra engineer with no skilled Python expertise. However when the identical job description is preceded by paragraph saying how my earlier roles makes me “excellent for this job”, it actively insults my intelligence.

What makes it further maddening is an LLM might have a look at my resume to see that I’m a match and write about how my expertise immediately ties in to the function. Or conversely it might decide “hey why the hell are we even speaking to this particular person, they do not know Verilog”. I think the latter is why they don’t really do that.

All of that coupled with even the job description being stuffed with extraordinarily frequent LLMisms pushes these emails beneath the plain previous templated emails in my thoughts. A easy edit move by way of one other LLM might take away these. Or, , doing a fast proofread of the factor being despatched out. The overt fakeness of the hassle and personalization makes it really feel so unapologetically impersonal.

#llms

#recruiting

#software





Source link