Shrey Agarwal

Late post: It was a pleasure to give an invited talk at Windsor University's CS colloquium/ seminar series and share our lab research on how we are leveraging LLMs in different phases of software engineering. The title and abstract are as follows, and the talk is available on Windsor University's YouTube channel. Leveraging Large Language Models in Software Development Lifecycle. Abstract: The advent of AI-driven technologies is revolutionizing software engineering, automating complex tasks like generating user story and code generation, program comprehension, bug detection, and maintenance. Among these advancements, large language models (LLMs), such as GPT, have emerged as transformative tools, enabling natural language-based code generation, refactoring, and documentation. These technologies not only enhance developer productivity but also reduce errors, improve decision-making, and accelerate development timelines. In this talk, I will explore how AI and LLMs are being leveraged to create innovative tools and techniques that streamline and enhance software engineering workflows. I will discuss the benefits these technologies bring to developers, as well as the challenges they introduce, including issues of scalability, explainability, trust, and ethics. By addressing these challenges, we can foster a collaborative future where human expertise and AI-driven solutions seamlessly integrate to produce reliable, efficient, and high-quality software. #seminar #iselab #InvitedTalk #uwindsor #usask

27

I received a PhD from MIT in 2022 in the field of Machine Learning. Since then, along with Rajat Dandekar and Sreedath Panat: I am on a mission to make AI accessible to all. As part of this mission, we started a series to teach Machine Learning. We called it "Machine Learning: Teach by Doing". We recorded 37 high quality videos and uploaded all of them on Youtube. All videos are available for free and are made with a lot of passion. The playlist has 100000+ views now. Here are the links for you to learn (shown for 30 videos): (1) Introduction to Machine Learning Teach by Doing: https://lnkd.in/gqN2PMX5 (2) What is Machine Learning? History of Machine Learning: https://lnkd.in/gvpNSAKh (3) Types of ML Models: https://lnkd.in/gSy2mChM (4) 6 steps of any ML project: https://lnkd.in/ggCGchPQ (5) Install Python and VSCode and run your first code: https://lnkd.in/gyic7J7b (6) Linear Classifiers Part 1: https://lnkd.in/gYdfD97D (7) Linear Classifiers Part 2: https://lnkd.in/gac_z-G8 (8) Jupyter Notebook, Numpy and Scikit-Learn: https://lnkd.in/gWRaC_tB (9) Running the Random Linear Classifier Algorithm in Python: https://lnkd.in/g5HacbFC (10) The oldest ML model - Perceptron: https://lnkd.in/gpce6uFt (11) Coding the Perceptron: https://lnkd.in/gmz-XjNK (12) Perceptron Convergence Theorem: https://lnkd.in/gmz-XjNK (13) Magic of features in Machine Learning: https://lnkd.in/gCeDRb3g (14) One hot encoding: https://lnkd.in/g3WfRQGQ (15) Logistic Regression Part 1: https://lnkd.in/gTgZAAZn (16) Cross Entropy Loss: https://lnkd.in/g3Ywg_2p (17) How gradient descent works: https://lnkd.in/gKBAsazF (18) Logistic Regression from scratch in Python: https://lnkd.in/g8iZh27P (19) Introduction to Regularization: https://lnkd.in/gjM9pVw2 (20) Implementing Regularization in Python: https://lnkd.in/gRnSK4v4 (21) Linear Regression Introduction: https://lnkd.in/gPYtSPJ9 (22) Ordinary Least Squares step by step implementation: https://lnkd.in/gnWQdgNy (23) Ridge regression fundamentals and intuition: https://lnkd.in/gE5M-CSM (24) Regression recap for interviews: https://lnkd.in/gNBWzzWv (25) Neural network architecture in 30 minutes: https://lnkd.in/g7qSrkxG (26) Backpropagation intuition: https://lnkd.in/gAmBARHm (27) Neural network activation functions: https://lnkd.in/gqrC3zDP (28) Momentum in gradient descent: https://lnkd.in/g3M4qhbP (29) Hands on neural network training in Python: https://lnkd.in/gz-fTBxs (30) Introduction to Convolutional Neural Networks (CNNs): https://lnkd.in/gpmuBm3j I have spent a lot of time and effort in making these lectures. I show everything on a whiteboard and then show it through Python code. Nothing is assumed. Everything is spelled out. P.S: We just launched a Generative AI Bootcamp curated by MIT PhDs. You can access the details here: https://lnkd.in/gnFn6Qq5 The next batch of this bootcamp begins on 5th April and we still have a few seats remaining!

4,506

141 Comments

Interesting... JEE prep, now via Google Gemini. 'New AI Tools to Support India’s Next Generation' https://lnkd.in/dFB_gWR4 (BLOG) "We’re launching full-length practice tests for JEE Main on Gemini, and soon in AI Mode in Search too. We have grounded the practice tests in rigorously vetted content from education companies like PhysicsWallah and Careers360, to build a best in class experience for learners. This helps ensure that you’re not just practicing — you’re preparing with material that more closely resembles what you’ll see on test day. To try it out, just tell Gemini “I want to take a JEE Main mock test”."

24

2 Comments

Received three reviews at AAAI on a paper that got rejected. AAAI introduced a novel reviewing mechanism to cope with the scale: each paper will receive three reviews in stage 1 -- two human-generated, one AI-generated. Unfortunately, we have strong suspicion that one of the "human reviews" is also AI-generated as some of the outlandish reasons the review cited to reject our paper just do not seem human. This means, the majority of the reviewers for our paper was AI. The human review, however, thought our paper was above the acceptance threshold. Let this sink in.

266

49 Comments

Most colleges are still “waiting” on AI agents. Sreenidhi’s SDC SNIST community chose to lead. Honoured to be invited as a guest to run a hands-on workshop on n8n Automations at Sreenidhi College. This is the first college I’ve seen step up and initiate a dedicated agents + automations workshop for students—and the response was incredible. Students were deeply engaged, asked sharp questions, and jumped straight into building. “Understood the basics of n8n and automations quickly—and built a few workflows during the workshop.” What we focused on: - How agentic workflows map cleanly in n8n (from trigger to deployment) - Turning repetitive tasks into hands-free automations - Clean, minimal stacks students can maintain without big budgets I’m grateful to the SDC community for the initiative and to every student who showed up to learn, experiment, and ship. It was a wonderful experience interacting with the Sreenidhi students—you made it special. Special thanks to Chandrashekhar M (Founder, SDC Community) for championing this initiative. — Master n8n automations — If you want to go from zero to deployed fast, join my Agents Master Course (n8n + real-world agent workflows). Practical, project-based, and built for outcomes: https://lnkd.in/gy_Ek-WD Prefer a quick strategy chat to automate your business tasks? Book a 30-min slot: https://lnkd.in/gj597xVS #n8n #Automation #AIAgents #NoCode #SDCCommunity #SNIST #FirstMover #EducationInnovation #StudentDevelopers #HyderabadTech #IndiaTech #Workshops #Repeatless #Btech #Colleges

523

20 Comments

Hello All, Here is a talk where I've expressed my perspective on how to place Indian knowledge systems in the modern day context of AI and technological advances.  Some of you may find it interesting -  https://lnkd.in/gbJiA_fw

305

13 Comments

I'm happy to announce "Agentic Patterns from Scratch" webinar. This free webinar will be jointly taught by me and Ajay Shenoy (PhD, IISc). During the webinar, we will teach how to implement agentic patterns from scratch. 2025 is considered as the year of AI agents. AI Agents are LLM based systems which can interact with environment, think and take actions to perform the given task. Agentic patterns are design strategies that dictate how AI agents think, act, collaborate, and improve. These patterns simulate human-like problem-solving and reasoning, offering flexibility and scalability over traditional rule-based systems. The four popular agentic patterns are - Reflection (self-critique and refinement), - Tool Use (interacting with external APIs or search), - ReAct (combining reasoning with actions), and - Multi-Agent (coordinating specialized agents). ➡️ Webinar details in the comments. #aiagents #agenticpatterns #learners #llms

374

23 Comments

Real-Time Object Detection for Smart Waste Bins – Powered by YOLOv5n Excited to share a smart little innovation from our Image Processing team at Fostride – this time, bringing ML into the world of waste management with Fostride’s Smart Bins! Yash Thakkar has built a real-time object detection system using YOLOv5n and OpenCV, trained on 3300+ images — and the results are genuinely impressive! 🔍 What it detects (very well): ✔️ Containers ✔️ Bags ✔️ Bottles ✔️ Wrappers This system will soon power Fostride’s AI-driven smart bins, helping sort and categorize waste more intelligently and efficiently. Just imagine — a bin that can see and understand what you’re throwing away. That’s the kind of real-world AI that makes a difference! Built entirely in Python, with seamless YOLO-style real-time labeling — fast, accurate, and lightweight thanks to YOLOv5n. The system outputs clean, annotated video that’s easy to integrate into our waste classification pipeline. 🛠️wTech Stack: Python, OpenCV, YOLOv5n 📁 Output: Real-time labeled video for smart bin integration We’re super excited about this step toward cleaner cities and smarter AI — one object detection at a time. Let’s build things that matter. #SmartBins #YOLOv5 #ComputerVision #Python #OpenCV #MachineLearning #IIClub #Fostride #AIForGood #ImageProcessing #WasteManagement #DeepLearning #RealTimeDetection #InternProject #TechForSustainability

570

103 Comments

🚀 सहज जीवन: probably the first open-source, collaborative, public, and live Marathi book, built transparently on GitHub. 🚀 The first part is a translation of Leo Babauta's "The Effortless Life" in Marathi, and the second part is open for contributions. I’ve already added a couple of chapters there myself. The idea is simple: anyone with useful insights about leading a good life can join and contribute. Contributions are voluntary, no money involved. 💡 Eligibility criteria for contributors: - Fluency in reading and writing Marathi - Basic GitHub literacy (fork, commit, PR). LaTeX expertise is not required - And yes, you must have read almost all of PuLa. That’s non-negotiable 😉 You can either refine the translation of the pending chapters or add fresh chapters of your own. Just fork the repo, add your changes, and send a PR. You can also  your add your name as co-author, thus build your own version of the book into a PDF, print it for yourself, and even publish it for wider circulation. I’m planning to self-publish a slightly updated version via Notion Press, which will naturally come with some costs. Feel free to do the same, the PDF is freely available in the repo. If you want to add a chapter, just place it at the end and submit a PR (Pull Request). If you are not comfortable with LaTeX or GitHub, you can share me the content via Google-Docs or email, I will add that on your behalf. 👉 If you’re interested in this whole “effort,” drop a comment below or message me your Gmail ID. I’ll schedule a short online meet or create a chat group to align next steps. IMO, this is not just a book, it’s a movement: bringing open-source culture to Marathi literature. Imagine being part of a collective that publishes a free, evolving, community-owned book! 📎 Repo name: yogeshhk/SahajJeevan (... on GitHub). 📑 Attaching the partial draft PDF here for a quick taste. (It’s a very rough draft, just to give you the idea.) जय महाराष्ट्र ✨

110

12 Comments

🎉 Recently wrapped up the “Build Rich‑Context AI Apps with Anthropic” course by DeepLearning.AI. It focused on the Model Context Protocol (MCP) — a practical approach to connecting LLMs with tools and data in a modular way. I found it helpful for thinking beyond just prompts and into how AI apps can be made more extensible, maintainable, and grounded in real context. If you're exploring how to make AI systems more useful and integrated, it's worth a look. #AI #LLM #Claude #Anthropic #MCP #AIApplications #DeepLearningAI #Chegg #CheggIndia #Edtech

45

2 Comments