Annkur retweeted
Hey @annkur, we need to print this out too, na? This is so epic from @tom_peters
1
2
4
555
RT @cnha: My wife is an animal rescuer and today, she got badly assaulted by some people in dommasandra village area
@BlrCityPolice we ar…
@BlrCityPolice we ar…
0
82
0
Annkur retweeted
Have you lost it @mybmc ? Please don’t do this.
🚨 Mumbai to get floating advertising barges and hoardings at waterfront sites across the city.
3
7
40
3,559
Annkur retweeted
The next Pune Knowledge on Tap event is this Thursday. Mandar will talk about Rebellion and Tradition in Indian Classical Music.
Registration link in next tweet
Registration link in next tweet
1
8
11
1,339
Annkur retweeted
This is good advice. Simple enough that you can start using it right away (meaning: it is not of the "how to manage 25 agents under $1000 per day" variety). But advanced enough that it will improve your outcomes.
Finally summarized what I've been sending to friends in DMs x.com/i/article/2114…
1
10
71
7,375
Annkur retweeted
Two hallmarks of Indian regulations in one place:
1. Can’t catch the people creating problems - let’s create pain for everyone. Why try to catch 1% when you can create the pain for 100%
2. More paperwork solves everything for the bureaucrats- it not really a solution but at least it’s not their problem anymore. Look at the problem that @rameshsrivats is talking about. RBI will say we published everything months ago, HDFC will say employee needs more training and that the software services company made an error, company will say this clarification is in their pipeline of fixes and that the clarification came too late, etc etc. A full round robin of finger pointing and no one who is responsible.
I am sure this behavior takes off a few bips off India’s GDP growth rate but that is problem for another ministry/department!
1. Can’t catch the people creating problems - let’s create pain for everyone. Why try to catch 1% when you can create the pain for 100%
2. More paperwork solves everything for the bureaucrats- it not really a solution but at least it’s not their problem anymore. Look at the problem that @rameshsrivats is talking about. RBI will say we published everything months ago, HDFC will say employee needs more training and that the software services company made an error, company will say this clarification is in their pipeline of fixes and that the clarification came too late, etc etc. A full round robin of finger pointing and no one who is responsible.
I am sure this behavior takes off a few bips off India’s GDP growth rate but that is problem for another ministry/department!
The EDF (Export Declaration Form) rules are now in effect, and it seems like RBI and banks have conspired to create a new hell. Or actually 5 hells.
I consult for an American company. Raised October invoice. Received the Inward remittance. And bang! Mail from @HDFC_Bank eeded before the payment can be processed.
Let me tell you what the 5 hells are:
1. The hell of confusion – RBI clarified last month that the EDF rules do not apply to individuals. But looks like they haven’t told the banks. And CAs also don’t seem to know. There’s no official communication on who is exempt. Last I checked, I was an individual. No company, no LLP. But HDFC still wants me to file the EDF to credit my payment.
2. The hell of complexity – See the attached pic of the form the bank has sent in excel. It has 5 sheets. To be filled and submitted. Some abstruse acronyms like “AD” - no clue what it means. Some fields are mandatory. Some fields are to be filled by the bank. No idea which is what. And it needs a physical signature. Which leads to...
3. The hell of paperwork - They also sent a pdf with a bunch of request letters. I’m supposed to choose the right letter, print, sign, scan and submit. Creating more paper today is incredibly silly. And, I suppose, someone in the bank will have to upload all this somewhere. Reminds me of those hand-filled immigration forms. Pointless.
4. The hell of repetition – At least for me, most of the details are constant. One client, one payment. If it’s a one-time exercise, fine. I’ll grit my teeth and get it done. But no. These details have to be given month after month. Else… no payment will be credited.
5. The hell of unpreparedness – It would have been simple enough for RBI or HDFC to create an online portal for this. With dropdowns, saved formats, contextual help, etc. to at least make this easy. But that would be too convenient, no? So, they’ve chosen to dump the entire burden on the taxpayer.
I have no opinion or understanding of why this is needed. Maybe it is critical information. Okay! But why make the process so confusing and painful?
And poor communication has meant that the banks have just passed on the pain to the exporter, and even CAs can’t help.
Basically RBI & HDFC feel I should consult less, and comply more. Maybe my client will also agree :)
17
72
292
10,239
Annkur retweeted
met a french woman in bali.
39, really tired, travelling alone.
we started talking at a cafe and she told me
she used to work in finance in paris.
made good money, visited over 40 countries,
never married, no kids, one very loyal croissant.
i asked if she ever wanted children.
she said yes, actually.
apparently she was with a guy for almost 8 years
but kept postponing the wedding and the kids
because she wanted to focus on her career.
then one morning her ovaries sent a calendar invite
titled "urgent: we need to talk" and she declined it.
she told me something interesting.
"at 28 i thought i had all the time in the world.
at 39 a stranger at a cafe asks me about my womb
before he asks my name."
i asked if she regretted travelling so much.
she said no, but she wished she'd spent more time
building relationships and less time
being interviewed by men who sell ebooks.
then she laughed and said her friends back home
are always complaining about their kids
while she wishes men would stop treating her uterus
like a quarterly earnings report.
ngl i found this really sad.
she had the money, freedom and passport stamps
everyone dreams about.
but the one thing i wanted now wasn't something
i could just book a flight and get.
engagement.
so i turned her uterus into a linkedin case study
and put the conversation in a gumroad ebook.
really nice woman though.
i wish her the best.
39, really tired, travelling alone.
we started talking at a cafe and she told me
she used to work in finance in paris.
made good money, visited over 40 countries,
never married, no kids, one very loyal croissant.
i asked if she ever wanted children.
she said yes, actually.
apparently she was with a guy for almost 8 years
but kept postponing the wedding and the kids
because she wanted to focus on her career.
then one morning her ovaries sent a calendar invite
titled "urgent: we need to talk" and she declined it.
she told me something interesting.
"at 28 i thought i had all the time in the world.
at 39 a stranger at a cafe asks me about my womb
before he asks my name."
i asked if she regretted travelling so much.
she said no, but she wished she'd spent more time
building relationships and less time
being interviewed by men who sell ebooks.
then she laughed and said her friends back home
are always complaining about their kids
while she wishes men would stop treating her uterus
like a quarterly earnings report.
ngl i found this really sad.
she had the money, freedom and passport stamps
everyone dreams about.
but the one thing i wanted now wasn't something
i could just book a flight and get.
engagement.
so i turned her uterus into a linkedin case study
and put the conversation in a gumroad ebook.
really nice woman though.
i wish her the best.
met a french woman in bali.
39, really pretty, travelling alone.
we started talking at a cafe and she told me she used to work in finance in paris.
made good money, travelled to over 40 countries, never married, no kids.
i asked if she ever wanted children.
she said yes, actually.
apparently she was with a guy for almost 8 years but kept postponing marriage and kids because she wanted to focus on her career.
they eventually broke up.
she told me something interesting.
“at 28 i thought i had all the time in the world. at 39 i don't feel that way anymore.”
i asked if she regretted travelling so much.
she said no, but she wished she'd spent more time building relationships.
then she laughed and said her friends back home are always complaining about their kids while she wishes she had one.
ngl that was a pretty sad conversation.
she had the money, freedom and passport stamps everyone dreams about.
but the one thing she wanted now wasn't something she could just book a flight and buy.
really nice woman though.
i wish her the best.
12
1
62
13,681
Annkur retweeted
eMudhra's ultimate homage to XKCD...
🤦🤦🤦🤦
🤦🤦🤦🤦
‼️ BREAKING: The private keys of 8 certificates from Indian certificate authority eMudhra were compromised, Mozilla says. Firefox blocked them within 3 hours.
Each key should be a 116-digit number, but one was just 4.
Anyone can guess them and forge HTTPS certs for any website.
5
56
633
11,542
Annkur retweeted
Lots of people are asking where the haze over Mumbai is coming from, and nobody seems to have a good answer. Let me try.
First, this isn’t surprising. October usually means an AQI of 110 to 140 in Mumbai, and that’s been true for most years since 2020, which is as far back as my t gets worse in November and peaks in January.
What you actually need to understand is where that high number comes from. US AQI is the air quality index used by the EPA, the US environmental agency. It looks at six pollutants. It turns each pollutant’s level into a score from 0 to 500 using a fixed table of cutoffs, and the highest of those scores becomes the AQI. The cutoffs are different for each pollutant, and they can change. The EPA made the PM2.5 cutoffs stricter in 2024.
India has its own index, the National AQI from the CPCB, and its cutoffs are bullshit. PM2.5 at 60 µg/m³ is “Satisfactory” on the Indian scale and “Unhealthy” on the US scale. Whatever app you use, check which AQI it’s showing you.
Now, why does Mumbai have haze, and why is the AQI so high? It’s PM2.5. Look at the orange line in Fig 2. PM2.5 is nasty stuff. The particles are fine enough to go deep into your lungs and fuck them up. It mostly comes from burning things, e.g., waste, vehicle exhaust, and industry. It’s mostly not from construction, which mainly produces the coarser PM10. Fig 3 shows PM10 falling and PM2.5 not falling. Last winter PM10 was the lowest it’s been, and the new construction dust rules probably helped with that. PM2.5 went up anyway.
The haze you see isn’t a good guide. Fine particles do scatter light, and humid, cooler air makes the haze thicker. But you can have heavy haze without much pollution, and you can have high PM2.5 with a clear sky. Trust a monitor, not the view out your window.
Sadly there’s no easy way to protect yourself from PM2.5. Your best bets are a well-fitted N95 mask outdoors and an air purifier at home. If someone is burning stuff near your place, that’s a big part of your problem, so shut the windows on that side.
Oh, and by the way, this probably isn’t because Mumbai has more cars. Fig 4 plots the growth in vehicles over the last few years against PM2.5. More cars have given us an NO2 problem, but they don’t explain the swings in particulate matter. The main cause seems to be somewhere else. Find out where things are being burned.
First, this isn’t surprising. October usually means an AQI of 110 to 140 in Mumbai, and that’s been true for most years since 2020, which is as far back as my t gets worse in November and peaks in January.
What you actually need to understand is where that high number comes from. US AQI is the air quality index used by the EPA, the US environmental agency. It looks at six pollutants. It turns each pollutant’s level into a score from 0 to 500 using a fixed table of cutoffs, and the highest of those scores becomes the AQI. The cutoffs are different for each pollutant, and they can change. The EPA made the PM2.5 cutoffs stricter in 2024.
India has its own index, the National AQI from the CPCB, and its cutoffs are bullshit. PM2.5 at 60 µg/m³ is “Satisfactory” on the Indian scale and “Unhealthy” on the US scale. Whatever app you use, check which AQI it’s showing you.
Now, why does Mumbai have haze, and why is the AQI so high? It’s PM2.5. Look at the orange line in Fig 2. PM2.5 is nasty stuff. The particles are fine enough to go deep into your lungs and fuck them up. It mostly comes from burning things, e.g., waste, vehicle exhaust, and industry. It’s mostly not from construction, which mainly produces the coarser PM10. Fig 3 shows PM10 falling and PM2.5 not falling. Last winter PM10 was the lowest it’s been, and the new construction dust rules probably helped with that. PM2.5 went up anyway.
The haze you see isn’t a good guide. Fine particles do scatter light, and humid, cooler air makes the haze thicker. But you can have heavy haze without much pollution, and you can have high PM2.5 with a clear sky. Trust a monitor, not the view out your window.
Sadly there’s no easy way to protect yourself from PM2.5. Your best bets are a well-fitted N95 mask outdoors and an air purifier at home. If someone is burning stuff near your place, that’s a big part of your problem, so shut the windows on that side.
Oh, and by the way, this probably isn’t because Mumbai has more cars. Fig 4 plots the growth in vehicles over the last few years against PM2.5. More cars have given us an NO2 problem, but they don’t explain the swings in particulate matter. The main cause seems to be somewhere else. Find out where things are being burned.
12
25
110
22,021
Annkur retweeted
My dad, age 93, can't read anything smaller than 1cm. So I completely replaced his Android's software with a customized app+launcher that does only the 5 things he needs in his phone, with large fonts (larger than the system's largest font in accessibility) and a simple interface
A question for you: what's the most useful (non-obvious) thing that you have AI agents do for you?
17
13
301
10,933
Annkur retweeted
Never did i think that i will be able to overlap Aviation x AI but that is exactly what we did!
I've always wanted a place to track all the planes i spot so we built Aerodex! Think Shazam but for airplanes!
apps.apple.com/us/app/aerodex…
As we set out to build it, we realised none of the existing models solved our problem so we ended up building our own! @Ayush_HRM_Singh has done a good job explaining that here
I've always wanted a place to track all the planes i spot so we built Aerodex! Think Shazam but for airplanes!
apps.apple.com/us/app/aerodex…
As we set out to build it, we realised none of the existing models solved our problem so we ended up building our own! @Ayush_HRM_Singh has done a good job explaining that here
1/8
🚀 RELEASING: Aerodex ID v7, an image classification model that predicts airline and aircraft typecode from a single photo.
Built by fine-tuning an ImageNet-pretrained ConvNeXt-Small.
Weights: huggingface.co/TerminalAero/a…
1
3
11
1,085
Annkur retweeted
Any connects at @tcl_india @TCL_Global need to get a replacement bill or challan for a TV replaced else @Zopper_IN is not giving extended warranty
0
1
0
126
Annkur retweeted
7/8
Aerodex ID v7 will soon power the aircraft recognition in our plane-spotter logbook app, Aerodex. The app has been live on the app store for two weeks.
Log your catches and give the model a try when it comes in the app:
apps.apple.com/us/app/aerodex…
Aerodex ID v7 will soon power the aircraft recognition in our plane-spotter logbook app, Aerodex. The app has been live on the app store for two weeks.
Log your catches and give the model a try when it comes in the app:
apps.apple.com/us/app/aerodex…
1
1
1
84
Annkur retweeted
5/8
Held-out evaluation: 93.7% airline accuracy and 95.4% exact typecode accuracy.
The test set included airframes absent from training. It can distinguish close variants, including the 737-800 vs. 737 MAX 8 and A350-900 vs. A350-1000.
Held-out evaluation: 93.7% airline accuracy and 95.4% exact typecode accuracy.
The test set included airframes absent from training. It can distinguish close variants, including the 737-800 vs. 737 MAX 8 and A350-900 vs. A350-1000.
1
1
1
72
Annkur retweeted
1/8
🚀 RELEASING: Aerodex ID v7, an image classification model that predicts airline and aircraft typecode from a single photo.
Built by fine-tuning an ImageNet-pretrained ConvNeXt-Small.
Weights: huggingface.co/TerminalAero/a…
🚀 RELEASING: Aerodex ID v7, an image classification model that predicts airline and aircraft typecode from a single photo.
Built by fine-tuning an ImageNet-pretrained ConvNeXt-Small.
Weights: huggingface.co/TerminalAero/a…
1
5
5
1,238
Another day, another poor LLM chatbot for customer support. This time from Urban Clap when cancelling a booking. At first I thought it is a post cancellation survey, but this is now the process to cancel!
What should be 2 clicks on an interface with charges seen is now a fight with an AI agent.
What should be 2 clicks on an interface with charges seen is now a fight with an AI agent.
2
0
5
665
Annkur retweeted
@annkur Ai ai PM and engineer everywhere ,not a single one to check their working product
0
1
2
64