Not sure why I am not getting any calls from HRs, even though my Naukri profile is getting viewed by multiple HRs, any suggestions?
#jobposting#Devops#Hiring
day 3 of building plug 🔌
it now finds who to ask.
first version returned the 3 highest-scoring people at a company. all three were basically the same person.
three peers is one bet placed three times.
so now it picks one of each:
peer — same function, likeliest to reply
senior — same function, referral carries weight
recruiter — paid to reply
different bets, not better and worse versions of one bet.
Day 2 of building my Job Referral Agent.
Today, I focused on turning a job description + resume into personalised referral outreach.
The goal: less generic AI spam, more relevant emails.
Terraform can pass while your AWS infrastructure is wrong.
Someone changes a security group manually. Your code and reality drift apart.
That's why IaC isn't just about provisioning.
It's about keeping infrastructure predictable, reviewable, and reproducible.
#Terraform #DevOps
@NamanyayG A big part of your product is keeping the LinkedIn/X accounts it drives from getting flagged. That's a reliability problem as much as a GTM one: proxies, pacing, warmup, idempotent sends. I run production Kubernetes infra for a living. Applying under "something else."
spent an hour on a send that kept failing with
"insufficient permissions" for a scope the code
didn't need.
the culprit:
getSetting_('TEST_INBOX', Session.getEffectiveUser().getEmail())
js evaluates default arguments eagerly. so that
Session call ran on every single send even though
TEST_INBOX was always set and the fallback never used.
one || later and the scope requirement vanished.
anyway. it sends emails now.
Applying to jobs is easy. Getting a reply is the hard part.
So I'm spending the next 7 days building my own AI agent that finds referrers, writes personalised outreach, and follows up automatically.
Free. Open source. Built with Claude Code.
Who wants to follow the build? 👇
We've gotten used to thinking:
AI = generate text.
Jev got me thinking about another possibility:
AI = make a decision.
Should I retry?
Which tool should I call?
Should this request proceed?
Sometimes an AI agent doesn't need another paragraph.
It just needs a decision.