About
“”
Within 2 weeks, if I’m unable to pick up any new technology your firm requires…
Services
Activity
2K followers
-
Haramrit Singh Khurana reposted thisHaramrit Singh Khurana reposted thisI've been a software engineer for over 7 years and today I build distributed systems at AWS. One key thing I've learned is this: I don’t worry too much if a junior engineer isn’t great at coding yet. What matters more is how they work: → Can they communicate clearly? → Do they collaborate well? → Do they take ownership? → Do they want to learn and grow? Coding is teachable. → AI tools, StackOverflow and YouTube can help. → Building projects and practicing DSA can help But being a team player? That's harder to teach. Technical skills are easy to develop. Behavioural skills are what truly set engineers apart.
-
Haramrit Singh Khurana reposted thisHaramrit Singh Khurana reposted thisHUGE: Official H1B lottery numbers were declared. This year, overall H1B registrations dropped 38%. Many more Indian students got selected! The numbers: → 211,605 registrations (down from 470,342 in 2025) → 50-60% selection rate (up from 29%) Best part: 71.5% of selected applicants hold a US master's degree or higher (up from 57%) What happened? The $100K H1B fee targeted outsourcing firms (TCS, Infosys, Cognizant) bringing workers from India worked well. These companies filed 200,000+ fewer registrations. Students on F1/OPT are exempt from this fee and so employers hired them more freely. Result: Less competition, better odds. Numbers also suggest 17% Level I applicants (fresh graduates) got selected this year which is quite decent. This is the best news Indian students have had in years. Trump’s policy designed to "crack down on immigration" accidentally created more space for students already in the US. Want the full deep dive? I’m covering it in today’s YouTube video (link in the first comment) Watch it to make well-informed decisions about your career and life. PS. Join 75,000 people who watch our weekly long analysis and tips on the YT channel (Scholar Strategy TV) if you haven't 🙏 #scholarstrategy #24northstar
-
Haramrit Singh Khurana reposted thisHaramrit Singh Khurana reposted thisInventory errors can quietly cost warehouses hundreds of thousands every year. Here’s how the best teams prevent it 👇 There's more to a warehouse than just loading and unloading. The real game is Inventory Control. One wrong stock movement leads to Expiry losses, Quality issues and Angry Customers. And it usually starts with a small mistake. Here are the principles that actually keep warehouses running right: 1️⃣ FIFO – First In, First Out What comes in first goes out first. A simple rule that prevents aging inventory. 2️⃣ FEFO – First Expiry, First Out Products closest to expiry move first. Critical in food, pharma, and cold chain. 3️⃣ Batch / Lot Control Every unit can be traced back to its source. When something goes wrong, you know exactly what to isolate. 4️⃣ Quality Segregation (Quarantine) Damaged stock is separated immediately. Not shipped, Not mixed., Not ignored. 5️⃣ System Driven Allocation (WMS) Strong warehouses do not rely on memory. Systems decide what gets picked and shipped. These aren't advanced concepts. They're basics that most operations claim to follow, but few actually do. The gap between a warehouse that bleeds money and one that runs clean usually comes down to one of these five. Which one does your operation struggle with most?
-
Haramrit Singh Khurana reposted thisHaramrit Singh Khurana reposted this$10 hardware. 10MB RAM. 1 second boot. That's a full AI assistant. Just found PicoClaw. While everyone's running AI on $599 Mac Minis, these folks are running it on a $9.90 LicheeRV Nano. The comparison: OpenClaw: TypeScript, >1GB RAM, >500s startup, $599 NanoBot: Python, >100MB RAM, >30s startup, ~$50 PicoClaw: Go, <10MB RAM, <1s startup, $10 99% less memory. 98% cheaper. 400x faster boot. Single binary. Runs on RISC-V, ARM, x86. The wildest part? 95% of the code was written by an AI agent. They used AI to bootstrap the entire Go migration. GitHub in comments.
-
Haramrit Singh Khurana reposted thisHaramrit Singh Khurana reposted thisNever leave your AI training unattended during the holidays. An OpenAI researcher forgot to stop a training run before vacation. When they returned, the model had undergone "grokking." The model switched from pure memorization to genuine understanding. This phase transition happens suddenly after extended training. Welch Labs breaks down this discovery in their latest video: • Models initially memorize training data perfectly • They show zero generalization to new problems • Then something clicks after thousands more steps • Sudden jump to real pattern recognition The researcher's accident revealed a critical insight. More training time can transform memorization into understanding. This explains why some models seem to "get it" unexpectedly. The grokking phenomenon changes how we think about AI learning.
-
Haramrit Singh Khurana reposted thisHaramrit Singh Khurana reposted this2026 is the year AI starts doing most of the shopping for us! Yesterday at National Retail Federation 2026, Sundar Pichai explained how AI is slowly taking over the entire shopping journey. And this wasn’t like any other another AI feature. Google introduced something much bigger: The Universal Commerce Protocol (UCP). AI isn’t just helping with one part of shopping anymore. It’s starting to handle everything end to end. From finding products and comparing options to checking availability, completing checkout, and even coordinating delivery. In simple terms, UCP lets retailers, payment systems, inventory, delivery partners, and AI agents speak the same language, so shopping can happen smoothly in the background without users jumping between apps or websites. Why does that matter to us? Because it means: → Less app hopping → AI can find what you need, check availability, compare prices, and buy it faster → Buy buttons can show up directly inside Google Search or AI tools like Gemini → Shopping becomes quicker, simpler, and more personalized This isn’t just talk. The protocol is already live and open. More details here: 👉 https://lnkd.in/g7idfrGE
-
Haramrit Singh Khurana reposted thisHaramrit Singh Khurana reposted thisClean Code tip: Prefix your booleans. Here's a simple cheatsheet I follow: • IS for simple states ❌ active ✅ isActive • HAS for ownership ❌ subscription ✅ hasSubscription • SHOULD for expected behavior ❌ retry ✅ shouldRetry • CAN for capabilities ❌ edit ✅ canEdit Your IF statements should read like plain English. Write code for humans to read, not just for machines to execute. PS.: To learn more Clean Code tips, check out my FREE newsletter: craftbettersoftware.com
-
Haramrit Singh Khurana reposted thisYes! There is hacking and there is software engineering. Sreya is doing good work here.Haramrit Singh Khurana reposted thisThis week, I worked on a ticket that seemed pretty simple “If the feature flag is ON, show components in one order. If it’s OFF, show them in another.” I opened the file… and found one massive component. Everything (logic, layout, and state living) is in the same place. At first, I just made it work. The logic passed. The ticket was ready for PR. But then I paused. It worked, yes but it wasn’t good. So I started breaking it down: 🧱 Split the page into smaller, reusable components. ⚙️ Moved the order logic into child components. 🔍 Noticed too many props being passed and my next goal: reduce prop drilling. By the end, the code wasn’t just functional it was maintainable. Readable. Scalable. Something the next engineer could easily pick up and make even better. And that’s when it hit me, Software engineering isn’t just about getting things to work. It’s about leaving things cleaner for the next person.
-
Haramrit Singh Khurana reposted thisHaramrit Singh Khurana reposted thisIntroverts: long load time, high value. Extroverts: instant popups, questionable ROI. Change my mind...
Haramrit Singh Khurana replied to a comment
1mo
Silas Stanley Thomas a combination of both would be ideal, but I def lean towards having a great mentor. They sometimes stick around even when you’re not colleagues anymore.
Haramrit Singh Khurana replied to a comment
1mo
Raman Walia you’re right. Context and being explicit matters a lot when working with AI models.
-
Haramrit Singh Khurana liked thisHaramrit Singh Khurana liked thisYou’re trying to pitch an object instead of diagnosing a pain. Buyers don’t care about water. They care about why they need it right now. If you ever struggle to “sell the water”, here’s the real problem. Here’s the practical reset: 1. Stop describing the product. Start naming the discomfort. Instead of “this is clean water”, say “You’ve been speaking for 2 hours. Your throat’s dry. Want to fix that in 3 seconds?” 2. Connect the offer to an immediate outcome. People buy relief, speed, clarity, certainty. Not features. 3. Anchor the gap. “What happens if you stay in this discomfort another hour?” Make the cost of inaction heavier than the price. 4. Close by removing their mental friction. “Here. One sip. If it doesn’t help, don’t take it.” Low resistance = fast decisions. 5. Remember the rule: The person who solves the pain fastest owns the deal. Stop selling the water. Start selling the moment it solves. 👉🏻 Follow Grant Cardone for more such amazing content 👉 Follow Sales Tips for no-fluff sales tips.
-
Haramrit Singh Khurana liked thisHaramrit Singh Khurana liked thisHe labeled it “scrap” and forgot about it for years. Then one day, while washing dishes, he noticed something strange: If he cut a smiley face into it, the sponge suddenly worked better than anything else on the market. It was a complete accident, but it went on to become the most recognizable sponge in America. The numbers today are staggering: - $926M+ in lifetime retail sales - $340M in revenue (2024 alone — up 54% YoY) - 160+ products across 257,000+ retail locations worldwide Now it’s one of the most successful Shark Tank products ever. A box of “scrap” foam became a billion-dollar cleaning empire. The lesson isn’t the material or the smiley face. The lesson is this: He used it differently than it was intended, and had the conviction to go all-in when it worked.
-
Haramrit Singh Khurana liked thisHaramrit Singh Khurana liked this“She Didn’t Just Sell—She Showed Up Like a Pro.” At just 13, Scarlett Cardone stepped into the sales arena where adults fold under pressure—and thrived. In a world of cold calls and closed doors, she didn’t hesitate. She picked up the phone and proved that confidence beats experience when belief is real. No fear. No stumbles. Just value delivered with poise. She listened. She adapted. She handled objections without flinching and ended every call with gratitude. But the real magic? It wasn’t about closing tickets—it was about opening possibility. Behind her was Grant—not micromanaging, but mentoring. Teaching her the rhythm of resilience, the math of momentum, and the mindset of mastery. Every “no” wasn’t rejection—it was repetition. Training. Growth. While most kids are playing games, she’s learning how to play for life. Not because she has to. Because she wants to. 💡 Lesson? Age doesn’t define success—effort and energy do. And in the school of business, the earlier you learn the reps, the sooner you learn the rewards. 👇 Would you put your kid on a sales call—or are you still underestimating what they’re capable of? #salesman #grantcardone #billionaire #mindset #entrepreneurship #ideas #sales #success
-
Haramrit Singh Khurana liked thisHaramrit Singh Khurana liked thisIt was the 3rd of January when the people from my hometown decided to felicitate me. But for what? For being the only software engineer in the town. And guess what? I was still in my 8th semester. I wasn't even a software engineer yet. I smiled through it, but deep down, I felt a little guilty. Everyone was so happy and proud of me for something I hadn't become yet. Still, I kept telling myself, "Maybe not today, but one day, I will be worthy of this." A few months later, I got selected at Newgen Software. I joined as a Software Engineer Trainee. From day one, the faculty members gave their best to make sure we understood everything. When my On Job Training(OJT) began, I was assigned to a real-time project. Like every fresher, I had a lot of questions, but my team never made me feel like any question was too small. They patiently helped me, guided me whenever I got stuck, and made my journey so much easier. Away from home in a completely new city, I also found people who made sure I never felt alone. And today, on the 1st of July, I will walk into Newgen as a Software Engineer. It's my first day as a full-time employee not just at Newgen, but in the corporate world itself. Thank you, Newgen, for believing in me and giving me this opportunity. This is just the beginning, and I'm excited for everything that's ahead.❤️
-
Haramrit Singh Khurana liked thisHaramrit Singh Khurana liked thisDay 15 of my DSA Journey – I solved the "Daily Temperatures" problem! Today, I tackled the Daily Temperatures problem from NeetCode/LeetCode. I began with the Brute Force approach to grasp the problem before optimizing my solution. My Approach: - Traversed each day's temperature. - For every day, checked all future days until I found a warmer temperature. - Stored the number of days to wait in the answer array. - If no warmer temperature existed, left the value as 0. Biggest Takeaway: This problem taught me how to search for the next greater element. While the brute-force approach is straightforward, it performs unnecessary comparisons and isn't efficient for large inputs. Complexity: - Time Complexity: O(n²) - Space Complexity: O(1) (excluding the output array) I know there's an optimized Monotonic Stack solution that solves this problem in O(n) time. If anyone has a clean explanation or implementation of the Monotonic Stack approach, I would appreciate your insights or code in the comments! Every problem teaches a new pattern, and today's lesson was understanding why optimization matters after first solving the problem correctly. Looking forward to learning the optimal approach! #DSA #Algorithms #Cpp #Coding #LeetCode #NeetCode #ProblemSolving #SoftwareEngineering #LearningInPublic #Tech #Stack #MonotonicStack NeetCode LeetCode
-
Haramrit Singh Khurana liked thisHaramrit Singh Khurana liked thisI studied how 6 legendary engineers debug code (so you don't have to): 0. 𝗝𝗼𝗵𝗻 𝗖𝗮𝗿𝗺𝗮𝗰𝗸 (methodical isolation): He doesn't guess. He strips the system down to the smallest failing case. Removes variables one by one. No intuition. Just process. 1. 𝗟𝗶𝗻𝘂𝘀 𝗧𝗼𝗿𝘃𝗮𝗹𝗱𝘀 (read the code, not the logs): While most devs scroll through stack traces, Linus reads the actual source. He believes most bugs come from wrong assumptions, not missing information. 2. 𝗠𝗮𝗿𝗴𝗮𝗿𝗲𝘁 𝗛𝗮𝗺𝗶𝗹𝘁𝗼𝗻 (defensive thinking): She didn't wait for bugs. She designed for them. Her Apollo code tested what should NOT happen. Every edge case was a real scenario. 3. 𝗚𝗿𝗮𝗰𝗲 𝗛𝗼𝗽𝗽𝗲𝗿 (find the actual bug): She literally coined the term "debugging." Her rule: go to the source. Don't trust second-hand reports. Reproduce the exact conditions yourself. 4. 𝗗𝗮𝗻 𝗔𝗯𝗿𝗮𝗺𝗼𝘃 (explain it to debug it): He writes down what the code should do vs. what it does. The gap between expectation and reality IS the bug. If you can't explain the fix, keep digging. 5. 𝗝𝗮𝗺𝗲𝘀 𝗚𝗼𝘀𝗹𝗶𝗻𝗴 (trace everything): He builds systems that explain themselves when they fail. Structured logs at every boundary. Correlation IDs across services. Errors that tell you exactly what went wrong. None of them start by guessing. None of them start by Googling the error. None of them start by asking ChatGPT. They all start by understanding. Method > talent. Pick one approach. Practice it for 30 days. Your debugging speed will double. What did I miss?
-
Haramrit Singh Khurana liked thisHaramrit Singh Khurana liked thisTwo senior engineers joined my team in the same month at Microsoft. Three years later, Nikhil was a Principal, and Akhil was still a Senior. I hired them both, so I had a front-row seat to the difference. On paper, they looked interchangeable: similar schools, similar years, comparable technical skills. For the first six months, I could not have told you who would pull ahead. The difference showed up in how they handled being wrong. Nikhil treated being wrong like information. When a decision of his turned out badly, he’d bring it to me before I noticed, walk me through what he’d missed, and adjust. It was almost boring how little drama came with it. Within a year, I was giving him decisions above his level. If one went sideways, I knew I’d hear about it early, and he would already be fixing it. Akhil was good, too. Sometimes very good. But he treated being wrong like an accusation. Every bad outcome arrived with a story attached. The data was late, the other team changed scope, the timeline was never realistic. Often the story was true. It also meant that over three years, I never once heard him say, “I got that wrong,” so I could never tell what he was learning. Here’s the part that took me years to see clearly. Akhil wasn’t lying. He genuinely believed each story, and that belief made it expensive. Promotions are bets, and I could price Nikhil’s mistakes. I knew how fast I’d hear about them and how quickly they’d get contained. Akhil’s mistakes had no price tag, because they came wrapped in explanations I’d have to unwrap myself. So when a Principal role opened, the decision was easy. You can coach someone who misjudged a situation. You cannot coach someone who was never wrong. If your explanations are always true and your career is standing still, those two facts are probably related. — Join 15000+ brilliant minds reading my weekly newsletter, The Conscious Leader, cultivating skills for modern leadership: https://lnkd.in/g3J25k5w
-
Haramrit Singh Khurana liked thisHaramrit Singh Khurana liked thisQuick tip if you're preparing for your next interview: Make good use of Claude's visualization feature to visualize your LeetCode solutions. It will help you learn better how the code works and also help you understand the core logic behind the whole solution. NeetCode also has this feature if you are a Pro member.
Experience
-
-
-
-
-
-
-
-
-
Education
-
George Mason University - Volgenau School of Engineering
-
Activities and Societies: PatriotHacks 2021
- Built an image classification app using PyTorch and Streamlit in less than 36 hours
-
-
-
-
-
Licenses & Certifications
Publications
-
A Resource Effective Approach for Distributed Machine Learning over a Local Network
IEEE 2019 International Conference on Advances in Computing, Communication and Control (ICAC3)
Areas of computer science such as data science and machine learning have boomed in recent years due to the increase in the availability of data. This has led to a need in the upsurge of computational power of various systems to handle large datasets consisting of images, text, audios, etc. However, training a machine learning model on a single node system can prove to be a tedious as well as time-consuming task. One of the solutions to this problem is distributing the model training process…
Areas of computer science such as data science and machine learning have boomed in recent years due to the increase in the availability of data. This has led to a need in the upsurge of computational power of various systems to handle large datasets consisting of images, text, audios, etc. However, training a machine learning model on a single node system can prove to be a tedious as well as time-consuming task. One of the solutions to this problem is distributing the model training process. Some of the current solutions that allow distributed machine learning are limited to multiple GPUs on a single node. This work attempts to propose a system which aims to solve this problem by providing a platform that allows distributing the machine learning process over a local network (using locally available nodes). This will ensure the effective use of all the resources available.
Other authors -
Courses
-
Analysis of Algorithms
CS 583
-
Component-based Software Development
SWE 645
-
Data Mining
CS 584
-
Database Systems
CS 550
-
Fundamentals of Systems Programming
CS 531
-
Mathematical Foundations of Computer Science
CS 530
-
Object Oriented Software Specification and Construction
SWE 619
-
User Interface Design and Development
SWE 632
Projects
-
Portfolio Website
- Present
See project• Designed a portfolio website using ReactJS
• Hosted on GitHub Pages with an auto-deploy setup with a push event as the trigger -
wha2do
- Present
- A minimal design todo list app
- Allows users to organize tasks into lists for effective management and tracking
- Implemented swipe gestures to make it easier for the user to mark task status
- Technology stack: React Native, FirebaseOther creators -
-
Logscribe-MCP
-
A lightweight Claude agent tool using Model Context Protocol (MCP) to analyze local log files
-
F1rstAid
-
-
MangoChat
-
See project- A simple chat application built on Node.js hosted using Express and Heroku
- Used NoSQL database systems to store message content, and socket.io for data exchange between clients and server
- Currently working on integrating third party authentication (GoogleAuth) -
CI/CD Project using AWS
-
• Deployed a sample student survey web app using Angular 8
• Used Docker for containerizing the application, Jenkins for continuous integration and deployment and GitHub for version control
• Deployed the Jenkins server and Kubernetes cluster on AWS EC2 instances assigning Elastic IPs to each instance
• Implemented the Kubernetes cluster using RancherOther creators -
-
Don Bosco Journal of Science & Engineering
-
Other creators -
-
SpaceTerra
-
See project• Created a single-player game using Phaser.io gaming framework for the front end, and Nodejs and MongoDB for the backend
• Hosted online for Teknack 2017, an annual gaming event organized by ACM DBIT
• Most played game at Teknack 2017 with over 20,000+ hits -
Macronux
-
• Replicated the underlying functionality of automating the process of starting up certain programs on bootup
• Developed for Ubuntu using Quickly GUI Maker, Python and GladeOther creators -
-
Sentimental Analysis using Twitter Movie Reviews
-
A system that fetches tweets based on a particular movie keyword and displays the percentage of positive tweets for that movie
Other creators -
Honors & Awards
-
2nd Place - Innovex Project Exhibition
Don Bosco Institute of Technology
- Won 2nd place at undergraduate school's annual project exhibition event for "Best Project Idea" under Department of Computer Engineering
- Also published a IEEE research paper on the project idea -
2nd Runner Up - Mumbai Hackathon ‘17
ERPNext and Don Bosco Institute of Technology
For the project “NumPy for C”
Test Scores
-
IELTS
Score: 8
-
TOEFL
Score: 105/120
-
GRE
Score: 319/340
Verbal: 154, Quants: 165, AW: 4.0
Organizations
-
Association for Computer Machinery (DBIT)
Technical Head
-- Organized Teknack, an institute-wide game development event, comprising of 15+ games developed by ACM-DBIT students, and the games played by over 50,000 players - Conducted an internal training session for 30+ student developers on web game development using Phaser.io, JavaScript, Node.js, and mongoDB
Recommendations received
4 people have recommended Haramrit Singh
Join now to viewOther similar profiles
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content