After giving 4 rounds of interview along with assessment with deployed url, the company reached out 4 weeks later for 3 more interviews.
Is it normal for startups?
Guardrails are integral part while dealing with AI, but it is not managed properly.
Every AI application rebuilds the same reliability layer repeatedly.
I started a project to make that unnecessary.
Checkout my new package - https://t.co/HEU2B79H9f
How do you evaluate that you have built a good system?
Follow these 4 steps:
1. Break system into components
2. Every component should have exclusive responsibilities
3. Technical detailing for each component
4. Each component is scalable, fault tolerant
The most useless feature ever built by @SlackHQ is split view.
If you wonder how, just click on it on the desktop.
Boom! half screen gone and empty.
This is from my experience after fighting for 1 hr to get it back to normal.
Not only features, change UX @SlackHQ
why limit yourself to someone else's preference and curation?
go build a db
go build a web framework
go write a file system
go build an operating system
go build a prog language
go write a webserver
go build an ORM
go build an agent
build whatever excites you
you have all the resources in the world to build literally anything you want. why follow a roadmap!
But what scares me is his attention to other things too, just before the incident he said he is forgetting names of people he rarely sees and can't work more than 1 hr without distraction.
I sense this as a serious problem with reels as I got to know other people like this.
Today one of my friends shared how his low attention span helped him to get out of depression
He is depressed with low appraisal at his work even after working Saturdays which they didn't compensate for. When a notification popped on his phone, he opened and went on to scroll.
At one point, he forgot what he was doing before. But I wonder how he was distracted. When I asked him about screen time, he was mostly scrolling reels in different apps.
Good thing is he got out of the moment and started working on upskilling.
After completing almost 2 years in Software industry, now I understand the proper use of 'I' and 'We' in group discussions will make you win in the company.
If loss functions tell an AI how wrong it is,
activation functions tell it how alive it should be.
They decide whether a neuron should “fire” or stay quiet - turning raw math into intelligent behavior.
Let’s break them down, one by one. (¬‿¬)
How to scale a system to support a million users:
• Keep web tier stateless
• Build redundancy at every tier
• Cache data as much as you can
• Support multiple data centers
• Host static assets in CDN
• Scale data tier by sharding
• Split tiers into individual services
Most people use neural networks.
Few truly understand how they see, think, and learn.
Let’s demystify it - from dense layers to convolutions, filters, strides, and latent spaces - in plain, intuitive language (¬‿¬)
This is how AI actually works under the hood ↓