Posts

The AI-native SDLC playbook · Claude Academy

linkedin tip: make genuine comments and network with people (not for the jobs) btw 9 to 6k here :)

thanks everyone :)

JevRelevance Open Source Released with 2.6x Precision

Launching 𝑱𝒆𝒗𝑹𝒆𝒍𝒆𝒗𝒂𝒏𝒄𝒆𝑹𝒆𝒕𝒓𝒊𝒆𝒗𝒆𝒓 and yes, it's open source :) $0.73 across 6,460 jev calls How to use: pip install jev-relevance You can: • filter retrieved chunks by relevance • configure thresholds and scoring modes • use per chunk or batched scoring • swap it into an existing LangChain RAG without changing your chain Built a separate evals repo to actually benchmark it. On the NFCorpus benchmark at the default 0.5 threshold, using BM25 candidate pool: • BM25 15.2% precision • Jev 39.7% precision (that's 2.6x the precision, while trading off some recall) Compared it against an LLM judge using the same candidate pool: • Jev $0.73 across 6,460 calls • LLM judge $0.71 across 323 calls Jev runs one call per candidate (20 × 323 = 6,460), while the judge runs one call per query. On average, each judge call uses ~12x the tokens and costs ~19x more per call. Added a dashboard with retrieval metrics, latency, per call costs and per query breakdowns. PyPI: https://lnkd.in/gcnf_2te GitHub: https://lnkd.in/gj-qPBNx Evals: https://lnkd.in/gVJCA7DE Dashboard: https://lnkd.in/gtbr2zxJ | 11 comments on LinkedIn

JEV: Fast, Accurate Decisions with Reinforcement Learning

Simulating Dice Throw with N Dice

I was revising LLD and came across a very simple Snakes & Ladders problem. Suppose there are N dice, and we need to simulate a dice throw and return the total. There are two ways to implement it. Only one is actually correct. 1. random number in the total range def makeTurn(self):   return r.randint(self.count, 6 * self.count) 2. roll each die independently def makeTurn(self):   return sum(self.rollOneDice() for _ in range(self.count)) Which one would you choose? Follow for more here: Saksham Malhotra

Back in December, I fixed a bug in npm. Small fix, but pretty cool to see it ship in a tool used by millions of developers PR: https://lnkd.in/dbDVYhfK #OpenSource #npm #SoftwareEngineering

Strengthening Backend Fundamentals with Notification System Design

It's always day one at Amazon! Grateful for the opportunity 🙏 Saksham Malhotra #InsideAmazon | 59 comments on LinkedIn

"50 DSA questions and CS fundamentals, that's all you need" ~ Myth The truth is, you can't expect to be below average in DSA interview preparation and get a decent or good job. Firstly, the number of DSA questions does not matter. What matters is whether you are able to solve two questions within 45 minutes in the interview environment. And let me tell you it's not an easy game. You should follow a sheet first and choose the one that provides the most clarity and covers each pattern effectively (Good news is that the patterns are limited.) You can start with tuf A2Z DSA, neetcode 150, etc. This is just the starting point. You have to solve company specific questions afterwards. You can find them on lc, geeksforgeeks, or even github and don't miss out on mock interviews. All you need to do is work hard, build consistency and discipline. Most importantly, have patience throughout this process Saksham Malhotra | 18 comments on LinkedIn

Small wins add up. Happy to share that I recently contributed to npm-cli and FreeCAD: - npm-cli: Resolved a Priority 2 bug, contributing to a tool that millions of developers rely on daily. - FreeCAD: Contributed two fixes to improve this open-source 3D CAD tool, which has a growing community of tens of thousands of users. Open source is a great reminder that consistent contributions can have a massive reach. Huge thanks to the maintainers for their time and thoughtful code reviews!

The days when my almirah became a whiteboard for DSA... Many of you have asked me how I prepared for my Amazon SDE Intern journey. I couldn’t respond to everyone individually so here’s the story It started with Amazon HackOn Season 5. My team, me and Dheeraj Sharma, worked tirelessly and made it to Round 2 after an assessment of coding questions and MCQs. Then came the idea submission round. We documented every detail, brainstormed every angle, and gave it our all. Yet we didn’t make it to the next stage. For a moment it felt like all the effort was in vain. But life has its own way of surprising you. A couple of weeks later I received an email. The very same HackOn performance that hadn’t carried us forward opened the door to one of my dream roles. That was my wake-up call. No shortcuts. No excuses. I had to ace the interviews. July and August became a battlefield. Revising DSA from takeUforward’s A2Z sheet. Understanding every concept behind each solution. Solving problems until patterns became instinct. But it wasn’t just DSA. My previous internships, development experience, and projects all came into play. LeetCode chart turned green day after day. My almirah-whiteboard stared back at me every morning leaving no escape from learning. Eventually all that grit and persistence paid off. One contest failure, one rejection, became the catalyst for one of the most exciting achievements of my career. Lessons I carry forward - Rejection is data not defeat - Deep understanding is more valuable than memorization - Consistency compounds - Every setback can trigger your next breakthrough For anyone feeling stuck your email is loading. Keep going (Attached: my almirah-whiteboard) | 36 comments on LinkedIn

My First day at Adobe I’m excited to begin one of the most meaningful journeys of my life — as a Summer Intern at Adobe Experience Cloud, within Adobe’s Digital Experience team. Honestly, I can’t put this feeling into words. I’m truly grateful, happy, and full of energy to learn, grow, and contribute in a place filled with such inspiring people. A big thank you to the amazing Adobe team for such a warm welcome and the smoothest onboarding experience, especially- Harpreet Kaur, Neha Kiroula, Anandi B, Samarth Arya and Ragini Khanna, for guiding us through a thoughtful and well-planned orientation. I’m also extremely thankful to Aayush Srivastava and Mrigesh Raj for opening the door to one of the greatest opportunities I hold today. Though this is just the beginning, I’m committed to giving my best, learning every day, and building meaningful solutions that make a difference. One more thing, special for the "my people" : A heartfelt thanks to my biggest supporters Poorvi Jain, Kanak Jain, Siddhant B., Anika Verma, Akshit Jain, Yash Singh and Dheeraj Sharma for standing by me throughout this journey. I wouldn’t be here without you. This day will stay with me forever. 𝕭𝖊 𝖔𝖇𝖘𝖊𝖘𝖘𝖊𝖉 ✍ #AdobeLife #AdobeIntern #Grateful | 55 comments on LinkedIn