Software engineer, teacher and a lifelong learner. Currently learning—AI Engineering. Interests—Education, AI & ML, Software Engineering, Applied Psychology.
🎙
So I met with @JayeshSidhwani and discussed engineering leadership at length. Jayesh was kind to come to my office and chat with me. We didn’t click pics to prove we met, but we managed to record it.
Do check it out. Feedback appreciated.
https://t.co/rmzEVUltsE
AI initiative or software roadmap, and not enough engineering bandwidth to move it forward?
That’s the delivery bottleneck we work with our clients to solve.
We'll be at GITEX Global in Dubai, December 7–11, and would love to meet businesses facing similar challenges.
I went on the @a16z podcast to talk about the opportunities in software and why I'm so excited.
Some points I make in the episode:
1. The interface is a means to an end. For business software, the value is in the work that gets done. AI assistants are becoming one more great way to reach that work, the way browsers were for the web. AI harnesses are the new browser.
I didn't say this in the episode, but wish I had: MCP (Model Context Protocol) is the new HTTP. If you aren't making your software product composable and callable by AI harnesses, you're going to miss this wave.
2. Don't use inference to solve something a query can solve. Structured data still matters. If you want to know who your top customers last quarter were, that should be a structured query with a deterministic response, not inference over a bunch of unstructured data presented as context to the LLM.
3. Don't use a frontier LLM to to do something a decision model (like JEV) can do.
4. AI is software. Really smart software, but software. It opens up a whole new class of problems to solve at every layer in the stack, so the opportunity is bigger than it's ever been.
What a time to be a building software!
Full episode here: https://t.co/1jZIPuphZw
Hi Twitter,
I have been duped by @ZeptoNow / @zeptocares , I got shipped a fake/stolen product and it felt like a genuine device when I received it this morning.
Upon using it for the day, I felt there was quality issues and upon trying to track warranty etc. I realised the product is not what was promised.
I am the same guy who went viral for ordering a Laptop via Flipkart Minutes, I have also ordered Google Pixel 10 earlier this year, so this is not my first rodeo ordering high net value items via Quick Commerce.
Am I posting here to get my money back? NO… read more to understand the context.
The nexus involved in this fraud is amazingly strong, they are able to infiltrate Zepto’s Supply Chain at some point.
Unfortunately, the call I did with Zepto Customer Care was not helpful. The staff was not competent enough to understand the issue, so here I am trying to get attention on this matter.
If I do not get support from them within 48 hours, I will offer a new AirPods 4 to a random person who retweets & replies to this tweet. Winners every Monday - every week for up to 4 weeks or until this issue is resolved.
Meaning, I am not doing this to get my money back, I am doing this to raise the issue. I am willing to burn more money than the Fake Airpods I received is worth, to prove that this is me trying to fight for attention to this matter.
Please tag media houses or Zepto Staff if you are aware of them, to help me garner attention.
I will add more details to this thread over the next few hours.
I'm now building an "improvements" feature into all of my personal vibecoded applications: a freeform text input that's hooked up to an agent that yolo implements whatever is submitted.
It's an interesting experience -- the activation energy barrier to sanding any rough edge or implementing any new feature becomes extremely low. Someone else who uses one of the applications I created has already implemented 150 improvements over the course of a few days. This pattern feels like a new kind of tool -- liquid software? -- and every application that doesn't have this feature now has a slightly rigid and anachronistic flavor. Why can't I just tell it how I want it to be different?
Now, it's not obvious to me how this pattern can or should scale. Does it become unmaintainable after a while? (Maybe model intelligence improvement outstrips my feature appetite such that it's fine?) How does collaborative software work in this model? Can software designed this way be secure and reliable?
Whatever the larger questions about broader deployment, it definitely works great in a personal context. It reminds me a little of the visions of Lisp/Lisp machines, Smalltalk, and HyperCard -- systems that the user can take control of and modify. Those visions never quite worked, though: the vast majority of software today is not end-user modifiable in any useful way. So, while this "liquid" affordance is a very small touch, the way it changes one's relation to the software feels a bit transgressive and empowering in a Lickliderian and Engelbartian way.
Startup engineering blog: "We chose PostgreSQL for its strict relational schema, type safety, and ACID guarantees."
Their actual database schema in production:
Our second product is called Blindsight and it will, as the name suggests, enable even those who are completely blind or have never seen to see.
It is ready from an engineering standpoint. Just waiting on regulatory approval.
Mumbai Twitter, I need your help 🙏
Looking for a 3 or 4 BHK on rent in/around Navi Mumbai, Thane, Seawoods, Kharghar, or Chembur.
Ideally in a modern gated society with good kids’ play areas and a temple. If you live in such a society, please check your society groups and DM me if anything is available.
Brokers, please excuse.
A trend I've been seeing lately in interviews is that engineers, especially those with less than a couple of years of experience, can't even code a basic implementation. They have delegated all their skills to LLMs.
If, as an engineer, you can't code, you are a glorified coding agent operator. And I won't hire you if that's all you bring to the table.