Shashwat Gupta 🚀 AI SaaS Founder

Founder @ConversAI_Labs • Ex-Nykaa • Building StoreCalendar—AI captions for Shopify 📈 #BuildInPublic

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

@MGSupportIndia I have sent you DM

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

@MGMotorIn Extremely disappointed with the lack of support.
₹23,999.99 paid on 23 Jan 2026 for MG Shield (Extended Warranty + RSA) for my MG Astor was successfully debited via UPI, but the plan is not activated and no response to emails.
Transaction ID: T2601231511571312531918

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

@matt_elevenlabs 11B

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

@reno_product_ai Hi
How can I reach out to you guys

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

@damianplayer AI

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

https://t.co/wVESeV4Ivt

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

Here are the key points from the story of Alex Krizhevsky and his role in the AI boom:

• In 2012, Alex Krizhevsky, a grad student at the University of Toronto, and his colleagues entered the ImageNet competition to evaluate AI algorithms for object detection and image classification.

• Krizhevsky's key insight was using GPUs to speed up deep neural network processing times, allowing him to train the network on millions of images in just days.

• The resulting neural network, AlexNet, won the competition by a huge margin, outperforming all other research labs and marking a turning point in AI.

• Tech giants quickly recognized the significance of Krizhevsky's work and began heavily investing in deep learning.

• After working at Google, Krizhevsky has now joined Dessa, a deep-learning startup, to continue pushing the boundaries of AI.

• Krizhevsky's impact on the field is immeasurable, with his seminal research paper being cited over 24,000 times.

#AI #DeepLearning #InnovationStory #TechPioneers #SiliconValley

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

The golden rule of management: do what it takes to get the best out of your team. In technical leadership as in any other leadership role, the best measure of your performance as a manager is the performance of the team itself. That means you should be thinking about and spending time doing everything necessary to help individual team members do their best work, both independently and collectively.

Coaching
Your principal role as a manager is to get the best out of the people on your team, so in many scenarios it's more appropriate to describe your role as that of a coach rather than a manager. A coach is somebody who is on your side, a source of wisdom and guidance to everyone on their team. A coach is quick to provide critical feedback, but also the first to celebrate and praise success.

Benefits of Overcommunication
There's nothing worse for an employee than feeling like their manager doesn't communicate enough with them. In the absence of information, it's a natural instinct to assume the worst-case scenario; a lack of information can also be a prime source of anxiety and confusion.

Two Types of Decisions:
In an annual letter to shareholders in 1997, Jeff Bezos outlined a framework for classifying decisions into Type 1 and Type 2 decisions. Type 1 decisions are not reversible, and should be thought through methodically, carefully, slowly, and with great deliberation and consultation, Bezos wrote. If you walk through a door and don't like what you see on the other side, you can't get back to where you were before. Which programming language you use, for example, is a Type 1 decision.

Type 2 decisions are the opposite. They can and should be made quickly by high-judgment individuals or small groups. Which exact shade of gray a button is can be a Type 2 decision, as it's easily changed down the line.

Bezos's advice, which I'll echo here, is that using Type 1 decision-making for Type 2 decisions leads to slowness and failure to experiment and innovate.

Most of your day-to-day technical decisions are Type 2 and are best made quickly and revisited or confirmed after you've collected more data via a prototype or MVP implementation. This is because the most expensive element of most startup technical decisions is the engineering team's time invested in the solution. If you deliberately constrain the time (and thus cost) invested into validating a reversible decision, you're out only a small bit. Most Type 2 technical decisions become irreversible only after you've invested considerable time and new engineering on top of them, so be rigorous about evaluating progress early on. When in doubt, make the decision to reverse early.

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

We talk about Huggingface and other platforms(anyscale/togetherai). But i really like @replicate . It has easy to use apis for open llms. True serverless setup.

I just checked one such model-> to convert english to hinglish . results are amazing:
https://t.co/nlDAgCG43H

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

Thread on the Evolution of GPUs and their Expanding Role in Computing 🖥️

1. Traditional Role of GPUs:🎮 Originally designed for graphics processing, GPUs excelled in gaming and 3D rendering. Their architecture is highly efficient for parallel tasks, a staple in graphics processing.
2. Emergence of GPGPU:💡 The parallel processing capabilities of GPUs were found to be useful beyond graphics. This led to General-Purpose computing on Graphics Processing Units (GPGPU), expanding their applications.

3. Role of Technologies like CUDA:🔧 NVIDIA's CUDA has been pivotal in transforming GPUs for general-purpose use. It allows developers to use familiar languages like C, C++, and Python for programming on GPUs.
4. Types of Tasks Suited for GPGPU:⚙️ GPGPU is ideal for tasks requiring parallel computation, such as deep learning, scientific simulations, and financial modeling.

5. GPU Architecture:🏗️ NVIDIA GPUs like the A100 and V100 have numerous cores designed for parallel processing, contrasting traditional CPU architecture.
6. SIMT Architecture:🔄 CUDA uses Single Instruction, Multiple Thread (SIMT) architecture, effective for parallelizable tasks.

7. Architecture and Specialization:🌐 The A100 and V100 GPUs focus on deep learning and high-performance computing, the T4 is energy-efficient for inference tasks, and TPUs are specialized for TensorFlow operations.

[Deedy @deedydas]

Winter is coming for Indian Masters students in the US this year.

~50,000-65,000 Indian Masters students in CS are going to graduate looking for tech jobs that *most* will NOT find.

This is the single largest international student subgroup in the US.

THREAD

🧵

1/7

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

Has ChatGPT rendered online evaluations obsolete?

Following up on a series of proctored programming tests, I found that candidates achieved full marks. However, upon inquiry, they revealed that they utilized Google Lens to copy the code and then pasted it into ChatGPT

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

https://t.co/6Og5g58wBp

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

@MGSupportIndia 9953053281

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

@MGMotorIn I booked a Astor at Safdarjung showroom Delhi.I specifically asked if BS6 phase 2 will come in Astor .They misled that it is only for diesel car . I cancelled the booking on March 11. Everyday I am following up with them with no success

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

Prompt engineering is the skill of articulating your thoughts in a clear and concise manner for an AI to understand and act upon. It requires eloquence, precision, and attention to detail. #AI #engineering #prompting

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

@mathemagic1an @Microsoft @langchain Wow, this is pretty cool! Can't wait to see the potential of the Visual ChatGPT in blending image and text interaction. Great to see @langchain implementation as well. Keep up the amazing work, Microsoft! #AI #MachineLearning

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

@mathemagic1an Agreed! AI-powered learning platforms like ChatGPT have revolutionized education accessibility and efficiency, allowing anyone to easily access quality information on any topic. It's time we start acknowledging the immense potential AI has to transform the way we learn and educat

[Shashwat Gupta 🚀 AI SaaS Founder @shashuec]

@doomie Wow! The development of GPT-4 sounds absolutely phenomenal! It's exciting to see how AI technology is advancing to operate on modalities that are still unknown to us. Can't wait to see the potential applications for "zero-shot modality learning"! #AI #GPT4 #futuretech