Lokesh Agarwal

From Astrologer's Whispers to Real Connections: Reclaiming Belonging in a Lonely Age Have you ever consulted an astrologer, their predictions swirling in your head, only to feel empty when your day takes a different turn? The emptiness might not be about the details themselves, but about seeking outside approval for your happiness. This same feeling can creep in even when you achieve success. Remember climbing the career ladder, landing that dream apartment, only to find yourself lost in a quiet emptiness? In India's bustling cities, it's a familiar, painful emotion. We're surrounded by people, yet there's a little loneliness at the edges of our lives. Thinking back to Mrs. Deshmukh, your friendly neighbour with the legendary Til Gur Barfi's, who brought the whole neighbourhood together with tea and friendly chats. Those evenings of shared stories and laughter, a web woven by simple human connection, feel like a distant memory in our super-connected yet broken world. We chase success with thumbs glued to screens, building online bridges that leave us feeling more alone than ever. The pressure to create a perfect online self-drowns out the whispers of our own desires. But here's the real deal, that shiny designation and fancy car can't fill the hole left by genuine connection. There's hope, though! We can rewrite this story. Let's ditch the small talk and chase deeper connections. Share your fears and worries, break free from feeling isolated in this sea of ambition. Remember the warmth of family gatherings, the joy of shared childhood games? Revive those small traditions. Pick up the phone for a real heart-to-heart conversation rather than What's Apping. Every act of connection, no matter how ordinary, how small, builds a more satisfying life. Imagine hearing back from an old friend, reliving an old laugh, and being surrounded by people who truly see us, where we belong. Isn't that what life is all about? Together, one conversation, one meal, and one act of kindness at a time, we can tear down the walls of loneliness. We can rewrite the narrative of modern society, where connection becomes the foundation of a life well lived. It's not too late, or is it? #HumanConnections #BelongingMatters #OvercomeLoneliness #RealConversations #MeaningfulRelationships #CommunityOverConnectivity #AuthenticLiving #SocialWellbeing #RedefiningSuccess #TraditionsRewoven #FightIsolation #VulnerableAndReal #KindnessRevolution #SmallActsBigImpact #ReconnectingHumanity #IndianSociety #UrbanLoneliness #DigitalDetox #QualityOverQuantity #ConnectionCures Photo by Tim Marshall on Unsplash

9

I am excited to share the latest blog, co-authored by myself, Junhwi Kim and Vyas Sekar, detailing how we managed to successfully localized and improved LLM by combining Fine-Tuning and RAG. https://lnkd.in/gKDkgGPq

70

3 Comments

A super clear, straight forward example of how to do cost effective RAG using LangChain for your document pipeline and Groq for compute resources. The code demonstrates how to speed up RAG and reduce costs using quantized embeddings (reducing 32 bit values to 8) Nice job Sahibpreet Singh! https://lnkd.in/gf5QMCKK

44

2 Comments

Can AI Understand the "Yeah, Right" Moments? Last week, while scrolling through some headlines about AI advancements, I started thinking about something that feels almost too human to replicate—sarcasm. Not the obvious kind that’s dripping with irony, but those subtle, everyday comments we all make that are more tone than content. Sarcasm, humor, and tone are deeply embedded in how we connect. And yet, AI? It still struggles with this kind of nuance. Picture this: You’re hit with a CAPTCHA that says, "Oh, perfect! Another Monday. Just what I needed!" and asks you to choose whether the statement is sarcastic or genuine. Easy for us, right? We know when someone’s rolling their eyes, even if we can't see them. But for AI? It’s not so simple. Let’s push it further: Imagine a coworker says, "Sure, let's have that critical meeting at 4:59 on a Friday." We immediately understand the frustration hiding behind those words, but can an AI catch that? Does it get the unspoken "this is a terrible idea" that’s baked into the comment? The answer is: not yet. Or what about this classic line? "Oh, great, another email asking for feedback. Just what I needed to make my day even better." Humans can pick up the sarcasm in a heartbeat—we’ve all been there. But AI would still be taking that at face value, thinking you really want to fill out that survey. As AI continues to advance, the question is: Can it ever truly get us in the way another human does? Detecting sarcasm isn’t just about parsing words, it’s about understanding the situation, the mood, the context—and sometimes even shared history. That level of emotional intelligence has been a uniquely human domain. The future of AI will depend on closing this gap. Right now, AI can analyze millions of data points and predict behaviors, but it can’t quite lean back in its chair, smirk, and say, "Oh, sure, another meeting. Just what I needed!" But what happens when it can? When machines develop the ability to sense not just the words but the meaning behind them? That’s when the real transformation happens. So, will AI ever fully detect sarcasm? I think it’s not a matter of if, but when. We’re already seeing AI models inch closer to understanding context and tone, but detecting the "Yeah, right" moments will require leaps in emotional intelligence. And when that day comes, maybe we’ll finally get AI that not only performs tasks but understands the intent behind them. Now, that will be interesting. Until then, I guess we’ll just keep waiting for AI to learn when we’re rolling our eyes.

2

We are pleased to announce that you can now achieve up to an aggregate 1.3 million I/O operations per second (IOPS) and 12 GB/s throughput per OCI compute instance with the OCI Block Volume Service. In addition, you can also attach up to 32 Ultra High Performance (UHP) volumes to a single OCI compute instance. This service update is particularly useful for high I/O workloads, such as OLTP, AI/ML, data warehousing and analytics, databases with large data sets and extensive complex processing. https://lnkd.in/gFN8nrtF

1

An open letter from Meta employees to Mark Zuckerberg and Meta's leadership. Very well articulated, I couldn't have done it myself better! I left Meta end of 2023 because of many things similar to what's mentioned in the letter, so I can definitely relate.

88

/*********Mentorship**************/ I was watching today Andy Jassy's video (https://lnkd.in/gRV2mGPr) on career advice. He shared one of the very important thing on Mentor's. Mentor's overall objective is too help you grow. They let you do things which you even think you can not do and then give support if needed. They give honest feedback on areas of growth.

1

Insightful explanation of how do go channels work behind the scenes. Its all about yielding the ptocessor for other goroutines. https://lnkd.in/gbyQ4yH9

1

Ah...the joys and pain of working with distributed systems/microservices. I am sure this parody will ring close to reality in many settings for software engineers all around :-) To my Product Managers and non-technical stakeholder friends, this is a must-watch as no amount of storytelling can help shine light on the complexity and implications of a microservices architecture, leading to unrealistic expectations or a lack of appreciation for the associated overhead in software development.

13

2 Comments

OpenSearch has consistently improved its hybrid query capabilities and performance by adding features such as post-filters, aggregations, and query parallelization. With the release of OpenSearch 2.16, we introduced sorting capabilities for hybrid queries, enabling applications to sort results by specific fields or attributes. This enhancement offers greater flexibility and enhances the relevance and usability of hybrid queries. Minal Shah Vamshi Vijay Nakkirtha Varun Jain Fanit Kolchina #opensearch #hybridsearch

24

The journey towards gRPC starts by understanding RPC conceptually. That's what this video by Arpit Bhayani is all about. https://lnkd.in/gaUGrGGP

1

I'm excited to share the second part of our multi-part series on LLM parameters! available on DZone 🚀 In this installment, we (Prince Bose, Tharakarama Reddy Yernapalli Sreenivasulu and I) delve into the "Top-P" parameter, also known as nucleus sampling, to understand how it influences the diversity and creativity of generated text. 🔍 What is Top-P? Top-P controls the diversity of a model's output by selecting words based on their cumulative probability. Low Top-P Values (e.g., 0.1): The model chooses from the most probable words, leading to safer but often repetitive responses. High Top-P Values (e.g., 0.9): The model considers a wider range of words, enhancing creativity but potentially deviating from the main topic. 💡 Why Does This Matter? Adjusting Top-P helps strike the right balance between creativity and coherence, depending on your content generation needs. 👉 Read the full article here https://lnkd.in/ggxEYTD9 #AI #MachineLearning #NaturalLanguageProcessing #LLM #TopP #ContentGeneration #GenerativeAI

12

I am thrilled to share a recent blog authored by one of our senior engineers at Upstox, Anjali Jain.In 'Beyond Encryption: Innovating Protection for Personally Identifiable Information (PII) in the Digital Age,' Anjali delves into our journey of safeguarding PII data, reflecting our ongoing commitment to maintaining the highest standards of data privacy and security. This blog highlights the challenges we faced and the strategic improvisations we made to overcome each one. Anjali has done a remarkable job of breaking down each iteration of our solution, detailing the associated metrics and their impact!Whether you're a fellow #engineer, a #cybersecurity enthusiast, or simply interested in learning more about the future of #dataprotection, this blog is a must-read.Dive into the article to explore how we are pushing the boundaries of what it means to protect personal information in the digital world

77

4 Comments

The scale we deal with, the complexity it brings in and the impact we create for millions of customers across the globe, are defining moments for our teams at Amazon India. Every day, our tech teams invent and deliver things that were never thought as possible! #HearFromBuilders at Amazon on how they solve real-world problems with technology. #NationalTechDay #ItsStillDay1 #AmazonIndia #TechAtAmazon

21

Thats the exact explanation I gave to a group of investors last week. (… it’s just Precision/Recall, attributes vectorization as “embedding”, K-Nearest Neighbor, User Search Intent, DAG, ML algorithms, etc all over again …). LOL and they asked what’s new this time around. And I was like not much, ML is still the backbone, and AI is the inference and actions based on the model. It works better when there is a feedback loop in a closed system. But the current state of the art like ChatGPT is based on scraping the Internet (garbage in, garbage out). Kinda like early search engines. What is a better analogy is that NVIDIA/Snowflake etc are selling tools, aka pickaxes to prospectors. The gold is up to you to find. And the opportunity right now is to monetize industry specific (healthcare, travel, financial services) clean data to build models with (aka a treasure map for the prospectors). From an enterprise perspective, what’s out there today aka generative AI isn’t very useful. Too much hype. To build useful engines will require an effort similar to search revolution 2 decades ago. From

16

3 Comments

Work tidbits - Living on the edge! Over a year back, I received my patent in the Knowledge Annotation space after 5 years of wait - it triggered me to join this new team.  The space was/still is AI Accelerators, within Cloud Compute. My team specifically was the ‘last mile’, productionize and land the AI accelerators for the top cloud customers.  The work would have been fine in the last 2+ decades of my career - but these times were different. Everyone and everything was in the AI/ML race. Before we(or others) could plan a new stack - something better was out there - sometimes open sourced! Before anyone could make the perf/$ cheaper - There was an update on new numbers available. There were negotiations on software lifecycles, in favor of velocity.   Customers were eager more than any other time. They were feeling the same stress - best LLM, best Stable diffusion, best MLPerf, best ML stack, best chips, best networking, best price! We did very well overall and that part will continue - despite living on the edge. Recently things shifted and I have gotten time to breathe(joined a sister team), also write my thoughts/tidbits - 1.) A friend, a mentor once said - Go far, not fast! Life is not a race and you have a long way to go! I do.  2.) Build resilience. You will see a lot of ups and downs. You don’t always have time to feel sorry for the down, or cheer for the ups! Keep your emotions in the middle spectrum.  3.) Find the true allies, and keep them close. They will tell you when you are wrong, or not - but most importantly they will stick with you, for long. I just ran into my peer from 9 years ago, 2 companies back in the hallway today!  4.) Recognize that some folks are indeed for the fast, maybe also far.. but mostly fast. Let them be! Take 2 steps back. Build resilience.  5.) Invest in your team - My reports, people are/were the team's biggest strength, and the primary reason to keep us going. We certainly don’t end here. The jobs will always be demanding, and that is by design. But the 'how' is on you..   My Amazon manager (now a VP@Microsoft) wrote me a handwritten note that said - Shivani, slow down and take the time to smell the flowers - I have been practicing something like this lately..  I am taking my learnings and my tribe along - and certainly aiming to go far.  Please take care of yourself, no matter what stage! ##resilience #livingontheedge #farnotfast #machinelearning #google #cloud

62

4 Comments

Let’s face it, LLMs aren’t without limitations. In this discussion, Varun Grover shares the benefits of a RAG framework and provides an overview of how it works as well as practical use cases. Watch now: https://vrt.as/43MNQQx #AI #GenAI #LLM #RAG

2

Let me say this one more time for those in the back: Chromium is neither Chrome nor Edge Gecko is not Firefox WebKit is most assuredly not Safari If you’re relying on an automated testing framework that uses those, and not the actual shipping browsers, you’re exposing yourself to risk. You may find the risk acceptable; I can’t answer that for you. Though I know I am tilting at proverbial windmills, I, for one, do not find that risk acceptable. But to ignore the risk or pretend it does not exist is willful ignorance at best and incompetent malfeasance at worst.

80

26 Comments