Thursday, July 3, 2025

The AI Revolution - Are You an Early Adopter, Follower, or Naysayer?

To See All Articles About: Layoffs Reports
To See All Interview Preparation Articles: Index For Interviews Preparation

Course Link

In May 2025, Microsoft laid off 7,000 employees, which was 3% of their total workforce. However, immediately after, they made an announcement that shook the corporate world: they are going to spend $80 billion this year on AI infrastructure. Not over the next 5 or 8 years, but $80 billion within 12 months on AI data centers. This news should be very important for all of us, because this is a sign of the changes that are coming in the next 5-10 years.

In this video, we will talk about perhaps the biggest transformation of our generation: Generative AI and Machine Learning. 

## What's Happening Around Generative AI and Why It's So Important

Research is throwing up a lot of statistics. Recently, I was reading a research report which said that AI could potentially remove 300 million jobs across the world, which is around 9-10% of the total jobs that are currently available. Because of this, many people are scared that jobs will be lost and AI will take our jobs. However, the same fear was present when computers arrived, when smartphones arrived, when industrialization happened, and when the internet arrived. The same fear is being felt today. But I, being a technologist and an optimist, believe that this absolute event in global history is also perhaps the biggest opportunity for all of us.

The World Economic Forum predicts that between 2023 and 2027, there will be a 40% increase in demand for AI and Machine Learning jobs. In fact, since 2019, when it felt like we didn't even know about AI, AI-related jobs have been increasing by an average of 21% annually, faster than most other jobs out there in the world.

However, whenever there is such rapid movement, demand is very high, and supply is limited. I still remember when I was in school and college, ITES (IT Enabled Services) was blowing up. All our college students were in some call center or undergoing some training, and the demand was through the roof. You could walk into any call center interview, and your job was guaranteed, with amazing perks and a good lifestyle. You would get accent training for foreign languages, pick-up services, work in very good offices, and working hours were conducive, allowing you to study and work. It was a completely different era for 5 to 10 years. Then IT services came, and the same thing happened: people started going onsite to the US, UK, and Australia, earning in dollars and spending in rupees. It was a complete transformation.

The same thing I can see happening in AI as well. The demand for jobs is so rapid that the supply cannot keep up. In fact, in the US alone, it's expected that AI jobs will reach 1.3 million opportunities, but skilled labor is only around 640,000, nearly half of the demand that is actually required. For India, I tried to find the same report and saw one that said by 2027, there will be a shortfall of 1 million jobs. This means there will be a requirement for 1 million AI jobs, but there won't be enough people if we don't start investing in skilling people right now. Microsoft's announcement is in that direction itself.

Rat Smith, a senior leader at Microsoft, mentioned in an interview that AI is like the electricity of our generation or our age. Just like in 1900, electricity transformed everything, leading to industrialization, the light bulb, and people working longer hours, it changed the way we interacted with machines, and our jobs also changed accordingly. The same thing is going to happen with AI, where AI is ready to do so much day-to-day work, administrative work, and run-of-the-mill tasks, so that we can elevate our work and be far more useful in what we do, as against wasting time doing things that a machine or an LLM can do today.

## The Three Types of Reactions to Technological Shifts

While researching for this video, I thought about technological shifts. In my time, I have seen two of them: one, which I will definitely say is the computer itself, because I was born in 1980, so I saw the computer revolution, at least to the point where we saw the internet become such a force and an enabler. And the second, I do believe, is the smartphone revolution in 2008-2009, when the iPhone was released, which also changed the industry so massively. So I have seen these two waves, and I see that AI is going to be perhaps the third wave, at least in my life.

Whenever such a wave comes, there are three types of reactions and three types of people:

1.  **Early Adopters:** These are the people who don't resist this change; they embrace it. They see that it's impossible for people not to use this, and it would be foolish to say that every person won't be using this tomorrow. It's almost like if I had said in 2005 that there would be a phone with which we could use the internet, and because of that, every person would have a computer in their pocket, and if you move in that direction, you will make brilliant careers, people would have laughed at me and said, 'What nonsense are you talking about? Nothing is going to happen. Let's just stay on our desktops, and we are happy there.' You would have missed the wave. But there were people who were like, 'We know what's happening. We can see what's happening in the US. The world is so connected now that news from there reaches here instantly. We can see what companies there are investing in.' I am telling you that Microsoft is going to spend $80 billion, and it's just one company, and it will be spent in just one year. So imagine how important AI will be for the entire technology world. So clearly, there is a direction.

Then I was looking at this data: how long did it take these platforms to reach 100 million users? Netflix took 18 years, Spotify took 11 years, Twitter took 5 years, Facebook took 4.5 years, Instagram took 2.5 years, YouTube reached 100 million users within 1.5 years, TikTok reached it within 9 months, and ChatGPT had 100 million users within two months. Its average user base as of April 2025 is around 800 million, nearly 1 billion people. This means one out of seven people is using ChatGPT, a software that can replace the work of so many people and, of course, make work easier. This is the power, and you cannot deny it. So early adopters see these things.

2.  **Followers:** What do followers do? They look at early adopters and say, 'These people are going there; let's go there too, because something is happening there.' I will give an example from my own life: I joined ISB when ISB was 5 years old, so I would call myself an early adopter. Today, someone who follows ISB after 20 years is a follower because they see that many people have gone to ISB, it's a very good school, and you get good money, etc., so they should go. So these are people who follow. It's not that they will not win, but it's possible that their outcome will be slightly less than that of the early adopters.

3.  **Naysayers:** These are the people who don't believe that anything like this is going to happen. Even today, I meet people who say, 'AI will not replace humans. Take it in writing, my friend, within 50 years, you will see fewer humans and more AI. Our world will be around AI, and that will not be a scary or a bad world to be in. It will actually be, in my opinion, a more efficient world to live in, so that we have time for all the things that we, as humans, should have time for.'

## The Call to Action: Become an Early Adopter

Why am I telling you all this? I am telling you this because I want you to become an early adopter. Being an early adopter doesn't mean that if you didn't use AI in 2021, 2022, or 2023, you are left behind. Now, it means that if you don't embrace this fully in the next 5 years, you are now going to either be a follower or a naysayer, the third category who will definitely be fired or laid off.

To become an early adopter, what do you need to do? You essentially have to get skilled. Skilling is the most important thing. Of course, you can learn on your own, stumble, make mistakes, and achieve all this, but the truth is that this field is changing so rapidly and dynamically that getting professional help as soon as possible will be a better way to skill yourself. And that's why SimplyLearn offers the Professional Certificate Program in Generative AI or Generative AI and Machine Learning. The good thing is that this curriculum is actually designed by the E&ICT Academy of IIT Guwahati, so it comes from a very elite perspective and a certification that holds weight.

It's an 11-month program, live, online, and interactive, so it's not self-paced where you learn on your own. And if you really see, which is where I spent time, what the learning path is, what things are covered, it actually covers everything that one needs to know about Generative AI and Machine Learning right now. I am talking about this program because SimplyLearn sponsored this video, but of course, the key is for you to recognize which course will be best for you when you want to step ahead and make the investment in your skilling around Generative AI and Machine Learning. In my experience and research, I found this course to be quite complete in what it covers, and of course, the backing it has from IIT Guwahati and the fact that it also comes from a recognized platform.

I would encourage you to check out the course, see if it fits both your requirements, your aptitude, your budget, and then make a call. You will get certificates from both IIT Guwahati and IBM, who have also partnered for this course. So, the industry certification by IBM, there are also masterclasses by experts, and AMA sessions and hackathons so that whatever you learn, you can actually apply.

## The Market Potential and a Personal Anecdote

The market size of Generative AI in 2022 was about $40 billion. In the next 10 years, it is expected to reach $1.3 trillion. That's an annual growth rate of 42%. If any investment is giving you a 42% annual growth rate, take it with your eyes closed. And in my head, this is the investment to make. If we talk about India, it is said that by 2025-2026, AI will have a positive impact of $100 billion on India's GDP.

I joined Twitter in 2007. But at that time, I didn't take it seriously; it seemed like a very complicated platform. Who uses it? What kind of people are there? What do they talk about? etc. But in 2009-2010, there was a discussion on Twitter about something that people, especially in the tech world, especially in Silicon Valley, became very interested in. And I remember hearing about Bitcoin for the first time around that time. I thought it was nonsense. Now, did I have ₹10,000-₹20,000 to invest in Bitcoin at that point in time? Yes, but I didn't. Why? Because it was a technological shift where I was not an early adopter. In fact, I would argue I was a late follower because I bought my first Bitcoin around 2014-2015, and I actually became serious around 2019-2020. Yes, I am not a naysayer, but the point I am trying to make is, if I had invested ₹10,000 in Bitcoin in 2010 after seeing and reading everything on Twitter as an early follower, do you know what its value would be today? ₹2,370 crore. And even if I had lost that ₹10,000, I would not be poor today. That's the way you have to think about Generative AI and Machine Learning today.

If you learn this, invest your time in it, and diligently try to improve your skills, you may have a completely different outcome in the next 5 to 10 years from what you can get today. But if, God forbid, for whatever reason, this whole Generative AI and Machine Learning hype doesn't pan out, you will still end up good. You won't lose anything; you won't be poor; you won't be lost in your life, because that's the power of being an early adopter. When you make a move, you end up learning something new that will set you off for life. The question is, how high will you go? You won't go down. So I will encourage you to take that leap forward, invest in learning about AI professionally from a skilled place, from a certified place, from a place of repute. And because this video is sponsored by SimplyLearn, I have presented you with one course option which you can evaluate, and I think it might suit your requirements very well.
Tags: Agentic AI,Artificial Intelligence,Generative AI,Large Language Models,Technology,Interview Preparation,

Sunday, June 29, 2025

30 Interview Questions on AI Engineering (from the chapter "Introduction to Building AI Applications with Foundation Models")

To See All Interview Preparation Articles: Index For Interviews Preparation

Generate 30 questions (10 easy, 10 medium, 10 complex) along with answers from this PDF

Here are 30 questions (10 Easy, 10 Medium, 10 Complex) based on the provided PDF content, along with their answers and page references:


Easy Questions

  1. What term describes AI models like GPT-4 and Gemini?

    • Answer: Foundation Models.

    • Page: 9

  2. What enables language models to scale into LLMs?

    • Answer: Self-supervision.

    • Page: 6

  3. What are the two main types of language models?

    • Answer: Masked language models and autoregressive language models.

    • Page: 4

  4. What is the basic unit of a language model?

    • Answer: Token.

    • Page: 3

  5. Name one common AI engineering technique for adapting models.

    • Answer: Prompt engineering, RAG, or finetuning.

    • Page: 11

  6. What is the most popular AI use case according to surveys?

    • Answer: Coding.

    • Page: 20

  7. What does "human-in-the-loop" mean?

    • Answer: Involving humans in AI decision-making processes.

    • Page: 31

  8. What metric measures the time to generate the first token?

    • Answer: TTFT (Time to First Token).

    • Page: 33

  9. Which company launched the code-completion tool GitHub Copilot?

    • Answer: GitHub (owned by Microsoft).

    • Page: 20

  10. What does LMM stand for?

    • Answer: Large Multimodal Model.

    • Page: 9


Medium Questions

  1. Why do language models use tokens instead of words or characters?

    • Answer: Tokens reduce vocabulary size, handle unknown words, and capture meaningful components (e.g., "cook" + "ing").

    • Page: 4

  2. How does self-supervision overcome data labeling bottlenecks?

    • Answer: It infers labels from input data (e.g., predicting next tokens in text), eliminating manual labeling costs.

    • Page: 6–7

  3. What distinguishes foundation models from traditional task-specific models?

    • Answer: Foundation models are general-purpose, multimodal, and adaptable to diverse tasks.

    • Page: 10

  4. What are the three factors enabling AI engineering's growth?

    • Answer: General-purpose AI capabilities, increased AI investments, and low entry barriers.

    • Page: 12–14

  5. How did the MIT study (2023) show ChatGPT impacted writing tasks?

    • Answer: Reduced time by 40%, increased output quality by 18%, and narrowed skill gaps between workers.

    • Page: 23

  6. What is the "Crawl-Walk-Run" framework for AI automation?

    • Answer:

      • Crawl: Human involvement mandatory.

      • Walk: AI interacts with internal employees.

      • Run: AI interacts directly with external users.

    • Page: 31

  7. Why are internal-facing AI applications (e.g., knowledge management) deployed faster than external-facing ones?

    • Answer: Lower risks (data privacy, compliance, failures) while building expertise.

    • Page: 19

  8. What challenge does AI's open-ended output pose for evaluation?

    • Answer: Lack of predefined ground truths makes measuring correctness difficult (e.g., for chatbots).

    • Page: 44

  9. How did prompt engineering affect Gemini's MMLU benchmark performance?

    • Answer: Using CoT@32 (32 examples) instead of 5-shot boosted Gemini Ultra from 83.7% to 90.04%.

    • Page: 45

  10. What are the three competitive advantages in AI startups?

    • Answer: Technology, data, and distribution.

    • Page: 32


Complex Questions

  1. Why do larger models require more training data?

    • Answer: Larger models have higher capacity to learn; more data maximizes performance (not efficiency).

    • Page: 8

  2. Explain how AI engineering workflows differ from traditional ML engineering.

    • Answer:

      • ML Engineering: Data → Model → Product.

      • AI Engineering: Product → Data → Model (due to pre-trained models enabling rapid iteration).

    • Page: 47 (Figure 1-16)

  3. What ethical concern arises from AI-generated SEO content farms?

    • Answer: Proliferation of low-quality, automated content risks degrading trust in online information.

    • Page: 24

  4. How did Goldman Sachs Research quantify AI investment growth by 2025?

    • Answer: $100B in the US and $200B globally.

    • Page: 13

  5. What inference optimization challenges exist for autoregressive models?

    • Answer: Sequential token generation causes high latency (e.g., 100 tokens take ~1 second at 10ms/token).

    • Page: 43

  6. Why might GPU vendor restrictions pose a "fatal" risk for AI products?

    • Answer: Bans on GPU sales (e.g., due to regulations) can cripple compute-dependent applications overnight.

    • Page: 35

  7. How does the "data flywheel" create a competitive moat for AI startups?

    • Answer: Early market entry gathers usage data → insights improve products → attracts more users → reinforces data advantage.

    • Page: 32 (Footnote 21)

  8. Contrast pre-training, finetuning, and post-training.

    • Answer:

      • Pre-training: Training from scratch (random weights).

      • Finetuning: Adapting a pre-trained model (application-level).

      • Post-training: Further training by model developers (e.g., instruction-tuning).

    • Page: 41–42

  9. What are the three layers of the AI engineering stack?

    • Answer:

      1. Application development (prompts, interfaces).

      2. Model development (training, datasets, optimization).

      3. Infrastructure (serving, compute, monitoring).

    • Page: 37

  10. How did the 2023 Eloundou et al. study measure "AI exposure" in occupations?

    • Answer: % of tasks where AI reduces completion time by ≥50%. Fully exposed jobs included mathematicians, tax preparers, and web designers.

    • Page: 17


These questions and answers comprehensively cover key concepts, use cases, and technical nuances from Chapter 1 of the provided text.

Questions generated using DeepSeek Tags: Agentic AI,Interview Preparation,Generative AI,Artificial Intelligence,Large Language Models,

Clocip KZ Soap (75g)

Index of Medicines
Clocip KZ Soap 75 gm

Composition
CETRIMIDE-0.5%W/W + KETOCONAZOLE-2%W/W

Manufacturer/Marketer
Cipla Health Ltd

Consume Type
TOPICAL

About Clocip KZ Soap 75 gm

Clocip KZ Soap 75 gm belongs to the class of antifungal agents. Soap is used to treat fungal infections of the skin. The lotion is used to treat dandruff. Fungal infection is a skin disease in which a fungus attacks the tissue and causes infection. Dandruff is a scalp condition characterized by dry, itchy, white flakes of skin on the scalp. 

Clocip KZ Soap 75 gm is a combination of two drugs, namely: Cetrimide (antiseptic) and Ketoconazole (antifungal). Cetrimide works by preventing the growth of microorganisms such as bacteria, viruses, and fungi. Ketoconazole works by damaging the fungal cell membranes that are essential for their survival. Together they help to treat fungal infections of the skin and also dandruff.

Clocip KZ Soap 75 gm is for external use only. Use Clocip KZ Soap 75 gm as advised by your doctor. Some people may experience skin irritation, dry skin, burning sensation, or skin rash. Most of these side effects of Clocip KZ Soap 75 gm do not require medical attention and gradually resolve over time. However, if the side effects persist or worsen, please consult your doctor. 

If you are allergic to Clocip KZ Soap 75 gm or any other medicines, please tell your doctor. If you are pregnant or a nursing mother, it is advised to consult a doctor before using Clocip KZ Soap 75 gm . Avoid scratching the scalp as it may worsen dandruff. Cover your head with a scarf or hat while going out. You are recommended to drink plenty of water as hydrating the body may reduce the number of flakes on the scalp, thus decreasing the appearance of dandruff.

Uses of Clocip KZ Soap 75 gm

Treatment of Fungal skin infections and dandruff

Directions for Use

Soap: Use it as advised by a doctor. Check the label for directions before use. Wet the body with water and rub soap gently to produce a rich lather. Leave it for few minutes and then wash thoroughly with water.Lotion: Use it as advised by a doctor. It is for use on the scalp only. Check the label for directions before using it.

Key Benefits

Clocip KZ Soap 75 gm contains Cetrimide and Ketoconazole used to treat fungal infections of the skin and dandruff. Cetrimide is an antiseptic that prevents the growth of micro-organisms such as bacteria, viruses and fungi. Ketoconazole is an antifungal that works by damaging the fungal cell membranes that are essential for their survival. Together they help to treat fungal infections of the skin and also dandruff. Thus, provides relief from a flaky and itchy scalp. 

How Clocip KZ Soap 75 gm Works

Clocip KZ Soap 75 gm is a combination of two drugs, namely: Cetrimide (antiseptic) and Ketoconazole (antifungal). Cetrimide works by preventing the growth of microorganisms such as bacteria, viruses, and fungi. Ketoconazole works by damaging the fungal cell membranes that are essential for their survival. Together they help to treat fungal infections of the skin and also dandruff.

What if I have taken an overdose of Clocip KZ Soap 75 gm

An overdose is unlikely due to low systemic absorption. However, seek medical attention if you experience any unusual symptoms.
Drug Warnings

If you are allergic to Clocip KZ Soap 75 gm or any other medicines, please tell your doctor. If you are pregnant or a nursing mother, it is advised to consult a doctor before using Clocip KZ Soap 75 gm . Clocip KZ Soap 75 gm is for external use only. Avoid contact of Clocip KZ Soap 75 gm with nose, mouth or eyes. In case Clocip KZ Soap 75 gm comes in contact with these areas accidentally, rinse with water thoroughly. Avoid scratching the scalp as it may worsen dandruff. Cover your head with a scarf or hat while going out. You are recommended to drink plenty of water as hydrating the body may reduce the number of flakes on the scalp, thus decreasing the appearance of dandruff. 

Diet & Lifestyle Advise

Regularly wash your hair with anti-dandruff shampoo. 

Use conditioners, shampoos and other products containing tea tree oil as they might help relieve itching.

Avoid using hair products containing harsh chemicals such as alcohol and bleach as they may dry out the scalp.

Avoid oily hair products that can build up on the scalp. 

Avoid sharing pillow covers with others and change them every 2 to 3 days.

Spend a few minutes in the sun every day as some evidence suggests that exposure to ultraviolet light can help control dandruff.

Manage stress with yoga, meditation, deep breathing and other relaxation techniques.

Habit Forming: No
Therapeutic Class: ANTIFUNGALS - TOPICAL

Wednesday, June 25, 2025

Agentic AI Books (Jun 2025)

Download Books


  1. AI Engineering by Chip Huyen
The LLM Engineering Handbook by Paul Iusztin and Maxime Labonne
Designing Machine Learning Systems by Chip Huyen
Building LLMs for Production by Louis-François Bouchard and Louie Peters
Build a Large Language Model (from Scratch) by Sebastian Raschka, PhD
Hands-On Large Language Models: Language Understanding and Generation by Jay Alammar and Maarten Grootendorst
Prompt Engineering for LLMs: The Art and Science of Building Large Language Model-Based Applications by John Berryman and Albert Ziegler
Building Agentic AI Systems: Create Intelligent, Autonomous AI Agents that can Reason, Plan, and Adapt by Anjanava Biswas and Wrick Talukdar
Prompt Engineering for Generative AI: Future-Proof Inputs for Reliable AI Outputs by James Phoenix and Mike Taylor
The AI Engineering Bible by Thomas R. Caldwell
Life 3.0 by Max Tegmark
AI Superpowers by Kai-Fu Lee
AI 2041 by Kai-Fu Lee
The Age of A.I. and Our Human Future by Henry Kissinger
Human Compatible by Stuart Russell
The Alignment Problem by Brian Christian
The Big Nine by Amy Webb
Creativity Code by Marcus du Sautoy
Competing in the Age of AI by Marco Iansiti
The MANIAC by Benjamin Labatut
The Art of Computer Programming by Donald Knuth
Artificial Intelligence: A Guide for Thinking Humans by Melanie Mitchell
Probabilistic machine learning by Kevin Murphy
Artificial Intelligence: A Modern Approach by Stuart Russell and Peter Norvig
Multiagent Systems: A Modern Approach to Distributed Artificial Intelligence by Gerhard Weiss
Artificial Intelligence: Foundations of Computational Agents by David L. Poole and Alan K. Mackworth
Intelligent Agents: Theory and Practice by Michael Wooldridge
Reasoning about Rational Agents by Michael Wooldridge
An Introduction to MultiAgent Systems by Michael Wooldridge
A Brief History of Artificial Intelligence: What It Is, Where We Are, and Where We Are Going by Michael Wooldridge
Build a Large Language Model (From Scratch) by Sebastian Raschka
LLM Engineer's Handbook: Master the art of engineering large language models from concept to production by Paul Iusztin and Maxime Labonne
AI Engineering: Building Applications with Foundation Models by Chip Huyen
Hands-On Large Language Models by Jay Alammar and Maarten Grootendorst
Why Machines Learn: The Elegant Math Behind Modern AI by Anil Ananthaswamy
Alice’s Adventures in a Differentiable Wonderland: A Primer on Designing Neural Networks by Simone Scardapane
The Little Book of Deep Learning by François Fleuret
The Basics of Reinforcement Learning from Human Feedback by Nathan Lambert
AI Snake Oil: What Artificial Intelligence Can Do, What It Can’t, and How to Tell the Difference by Arvind Narayanan and Sayash Kapoor
Co-Intelligence: Living and Working with AI by Ethan Mollick
Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence by Kate Crawford
Genesis: Artificial Intelligence, Hope, and the Human Spirit by Henry Kissinger, Eric Schmidt, and Craig Mundie
The Singularity Is Nearer: When We Merge with AI by Ray Kurzweil
Dancing with Qubits - 2nd edition: From qubits to algorithms, embark on the quantum computing journey shaping our future by Robert S Sutor
Nexus: A Brief History of Information Networks from the Stone Age to AI by Yuval Harari

Book Title: Building Agentic AI Systems
Authors: Anjanava Biswas, Wrick Talukdar, Matthew R. Scott, Dr. Alex Acero

"Superintelligence: Paths, Dangers, Strategies" by Nick Bostrom
"Artificial Intelligence: A Guide to Intelligent Systems" by Michael Negnevitsky
"Life 3.0: Being Human in the Age of Artificial Intelligence" by Max Tegmark
"Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow" by Aurélien Géron
"The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World" by Pedro Domingos
Tags: List of Books,Agentic AI,Generative AI,Technology,

What this book talks about - AI Engineering by Chip Huyen

Download Book

“AI Engineering” by Chip Huyen is a comprehensive guide to building real-world applications using modern foundation models (like GPT, Claude, Stable Diffusion), rather than training ML models from scratch github.com+15oreilly.com+15iseoai.com+15.


🧠 What the book covers

  1. Defining AI Engineering

    • Explains how AI engineering differs from traditional ML engineering by focusing on model adaptation—prompt engineering, retrieval-augmented generation (RAG), fine-tuning, agents—instead of pure model training iseoai.com+7mlops.systems+7barnesandnoble.com+7.

  2. The New AI Stack

  3. Planning AI Applications

  4. Adaptation Techniques

  5. Evaluation Methods

    • Discusses the challenges of evaluating open-ended LLM outputs

    • Introduces “AI-as-a-judge”—using AI to evaluate AI outputs—and the importance of robust metrics for dangerous failure modes mlops.systems+6oreilly.com+6tertulia.com+6

  6. Inference & Deployment Optimization

    • Defines latency/throughput metrics (e.g., time to first token, time per token)

    • Describes model-level (quantization, distillation) and serving-level (batching, caching, attention optimization) techniques reddit.com+3github.com+3reddit.com+3.


🧩 Who it’s for

  • Engineers, technical product managers, and startup founders building AI-powered applications

  • Those who want a product-first workflow: build with APIs early, then iterate with data and fine-tuning iseoai.comhowtoes.blog+1iseoai.com+1

  • Anyone seeking a hands-on roadmap: from selecting models/datasets & crafting prompts to optimizing inference and deployment barnesandnoble.com


✔️ Key Takeaways

Focus AreaInsight
Mindset shiftFrom traditional ML to AI engineering oriented around adaptation and evaluation
Techniques coveredPrompt engineering, RAG, fine-tuning, agents, quantization, caching
Evaluation focusHandling open-ended outputs and preventing “catastrophic failures”
Operational strategyLatency/cost trade-offs and optimization in deployment environments

📌 Summary

Chip Huyen’s AI Engineering (published December 2024 / Jan 2025) is a seminal manual for today’s AI practitioners. It walks you through the full lifecycle: from planning and developing AI apps using foundation models, through rigorous evaluation and fine-tuning, to real-world deployment optimized for performance and cost.

Whether you're a seasoned ML engineer transitioning into LLM-powered systems or a full-stack dev looking to integrate AI into products, this book gives you the framework, tools, and practical strategies to build robust, valuable AI applications.

Tags: Technology,Agentic AI,Generative AI,Book Summary,

Sunday, June 22, 2025

December 2012 (as remembered in Jun 2025)


Other Journaling Days

December 2012, the craziest month of the craziest year of my life.

On a side note: In the last edit (which Gemini refused to do) was to "change the setting to high rise residential society" :D What happened in the last fortnight of 2012 was a build up of what I was going through the rest of the year. I mean I had no problem with the surveillance even if it was just my imagination. I did not see any harm. Surveillance was on and I was reminded of it in the following sequence roughly with passage of time: 1. At first, it was people on public buses that I used to take for college 2. Fabrication of people, events around me like the times when I would catch a glimpse of any teacher anywhere on campus. 3. Molding and folding of situations when I was in internship after completing my seventh semester. Like when my summer internship at HCL Infotech was extended to run into my semester days and I had paid for Rs 10K for it. And then getting the certificate from them happened in December something that was to happen last by August. 4. Involvement of society friends like Hardik, Drishesh (Appu) into evening meet ups. Thinking of Hardik’s involvement still hurts. 5. Involvement of Mahima (with Ojas) and Naina Dargan (with Hardik) in the situation. I could make the above points a subject in their own right. Well, what happened in December is what a psychologist might call “mania” – a period of extreme highs. Hellish lows were to come by the January of 2013. In the last fortnight of 2012, I was singing songs. No, actually just one song repeatedly every evening. “The Rape Song”. Written by me. I wonder what it might have sounded like, what it might have appeared like to the people who were an observer to all that nonsense :D I mean, it was a crazy time. Cuz rapists rape They love it when you hate You show them hate, they show you disgrace Cuz rapist rape, cuz rapists rape…. Kind of like a 50 Cent and Eminem song with 50 at the chorus and Em’ on the rap. I strongly feel that had I known meditation back at that time then things would have been totally different. Part of my behavior could be clearly explained by inspirations from American artists, American culture and American media (Rap and Hip-Hop culture). Today, I observe that “actually” no one complained of my behavior explicitly or implicitly. It was just conclusions that I was supposed to build in my head like my other thoughts. And use my behavior from December as the reason for intense hammering that happened in February 2013 by Mudita Gupta, B-27, Manu Apartments (Mayur Vihar, Ph 1, Delhi).

Saturday, June 21, 2025

Power of Rs 1,50,000 PPF Investment... In Generating a Stable Income


Lessons in Investing
On a side note:

If you want to do this math yourself in an excel sheet, do check this Excel sheet out.

How many years will it take to generate Rs 85,000/month tax-free income from Public Provident Fund?

To earn Rs 85,000 per month from a PPF, one must start with an investment of Rs 1.50 lakh every year and continue it until the 15-year maturity period is reached. Later, you can extend the account for blocks of 5 years each to reach your target corpus. Let’s understand it with the help of calculations. The PPF is a government-backed investment scheme, which makes it secure to invest in. The returns are assured by the Government of India, providing a reliable option for investors, but remember you can hold a PPF account in only one individual's name. The Public Provident Fund also enjoys an Exempt-Exempt-Exempt (EEE) tax status. Which means, investments of up to Rs 1.5 lakh annually are deductible from your taxable income, the interest earned on the PPF is tax-free, and the maturity proceeds, including the principal and interest, are also exempt from taxation. Go through the article to figure out how you can get Rs 85,000/month tax-free income from Public Provident Fund. 1/14: What is Public Provident Fund? As mentioned above, Public Provident Fund, which is popularly known as PPF, is a long-term, government-backed savings scheme in India. It is designed to encourage consistent savings for the golden years. The PPF scheme has a fixed tenure of 15 years, with the option to further extend it in blocks of 5 years. 2/14: What are advantages of PPF? It offers guaranteed returns. Public Provident Fund is open to all individuals, including those who are employed or self-employed. Parents or guardians can open a PPF account for minors. 3/14: What are withdrawal rules in PPF account before maturity period? You can make one withdrawal per year after completing 5 years from the date of account opening. Note that the 5-year lock-in period includes the year of account opening. For example, if you opened your PPF account in 2024-25, you can make your first withdrawal in 2030-31 or later. 4/14: How much can you withdraw from PPF? When making a withdrawal from your Public Provident Fund (PPF) account, there are specific limits to keep in mind: 5/14: Public Provident Fund withdrawal limit You can withdraw up to 50 per cent of the balance at the end of the 4th preceding year or the end of the preceding year, whichever is lower. For example, if you are making a withdrawal in the financial year 2024-25, you can withdraw up to 50 per cent of the balance as of March 31, 2023, or March 31, 2024, whichever is lower. 6/14: What happens to your PPF account after 15 years? After completing the initial 15-year maturity period, you have the flexibility to manage your Public Provident Fund (PPF) account as follows: You can choose to continue your account with or without making further deposits. This allows you to extend the benefits of your PPF account beyond the initial maturity period. 7/14: How to get Rs 85,000/month income from PPF? To generate Rs 85,000 per month from a PPF, you have to begin with an investment of Rs 1.50 lakh every year and continue it until the 15-year maturity period. Later, you can extend the account for blocks of 5 years each for maximum return. 8/14: How many years will it take to build Rs 85,000/month tax free income? It will approximately take 29 years to reach this target corpus. 9/14: What will be PPF corpus after 15 years? The investment amount in 15 years will be Rs 22,50,000, the estimated interest will be Rs 18,18,209, and the estimated maturity will be Rs 40,68,209. The investor can take an extension of 5 years and keep investing Rs 1.50 lakh a year in the same way as before. 10/14: What will be PPF corpus after 20 years? In 20 years, the total investment will be Rs 30,00,000, the estimated interest will be Rs 36,58,288, and the estimated corpus will be Rs 66,58,288. At this stage, the investor can take another extension of 5 years and continue the practice of investing Rs 1.50 lakh a year. 11/14: What will be PPF corpus after 25 years? In 25 years, the total investment will be Rs 37,50,000, the estimated interest will be Rs 65,58,015, and the estimated corpus will be Rs 1,03,08,015. 12/14: What will be PPF corpus after 29 years? In 29 years, the total investment will be Rs 99,26,621, the estimated interest amount will be Rs 99,26,621, and the estimated corpus will be Rs 1,42,76,621. 13/14: What is next step after 29 years of investment? From here onwards, account holders can start withdrawing interest on the entire corpus. During extensions, the account holder is allowed to withdraw the interest amount once a year. 14/14: What will be your interest amount? At a 7.1 per cent interest rate, the interest in a year will be Rs 11,89,718, which will be equal to Rs 85,000 a month. Ref
Tags: Investment,Finance,