I am interviewing a lot of candidates these days
for forward deployed engineering roles
I am only looking for these 2 signals
in their story:
A] Sharp B] High agency
FDE question:
What is the practical way to purge duplicate content from semantic search, when you are not the SME of the department?
ex- an HR bot has 2.5K faqs questions ingested.
Some policies are duplicate
how do you keep search fresh?
It’s just been 18months and look where we are
There's a new kind of coding I call "vibe coding", where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. It's possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper so I barely even touch the keyboard. I ask for the dumbest things like "decrease the padding on the sidebar by half" because I'm too lazy to find it. I "Accept All" always, I don't read the diffs anymore. When I get error messages I just copy paste them in with no comment, usually that fixes it. The code grows beyond my usual comprehension, I'd have to really read through it for a while. Sometimes the LLMs can't fix a bug so I just work around it or ask for random changes until it goes away. It's not too bad for throwaway weekend projects, but still quite amusing. I'm building a project or webapp, but it's not really coding - I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.
Replying to @spadjay
That’s 90% of the orgs of the world.
Sadly there board hires people like me.
Top answer
xitter.gq/wayne_liang_/status/21…
Replying to @code_sagar
half of that is just waiting in line. stream tts per sentence, and pull the billing data the moment the call connects, you know the account way before they finish asking. the 700ms silence wait can drop a lot with a turn detection model too
Replying to @1Muhammad_AQIB
There is a caveat for starting the billing api call immediately.
Can you think about that?
Replying to @wayne_liang_
can you double click on the turn detection model ?
Forward deployed engineer question:
A UK energy retailer's voice agent takes 2.8s (p95) to reply.
The time goes to:
700ms waiting for silence
300ms transcription
600ms to first token
1.2s on a billing API,
then TTS synthesises
the whole reply before playing
Optimise this?
I went to Hindi medium govt school, wore used shirts and my college was funded by my relatives pooled money
life wasn’t easy
10 years after my graduation in 2014
my bloodline will never face
the horrors that I faced.
we as a family are now confused
on where to spend the year end
hope this helps put this in perspective of what this means.
this is a great initiative, love it.
but have you visited BLR KIA T1 on Monday 6AM and 10PM
you will see daily passengers like me
waiting in long queues in formals
few more flights would help definitely.
I booked a one way train ticket to Bangalore in 2014
with no job and 14K
Since then life has sky rocketed
met my wife
built sports team
built a stable career
bought real estate
got to work at
bleeding edge of AI
Thank you people of Karnataka.
My family is grateful 🙏








