Neeraj Agrawal

Hot off the press just minutes ago, Confluent just announced Freight clusters in #KafkaSummit Bangalore. They’re a new innovative cluster type that lowers prices by up to 90% (!) 🔥 What is it and how does it work? It’s all about latency. Some workloads like: 🔷telemetry 🔷metrics 🔷logs don’t care whether the latency is 50ms, 500ms or 2s. But the same workloads typically have a TON of data. And when you need to replicate that data 3x and pipe it through the network… the network costs can eat you alive 💀 So what did #CFLT do? They decided to write the data straight to S3! Instead of having Kafka replicate it across brokers, they just write it directly to S3. S3 itself does the replication under the hood (for much cheaper) 💸 The catch? S3 is much slower. But as we mentioned, those workloads don’t care. Similar to a cargo ship, you’re fine letting the cargo (data) take longer to ship as long as the price is good. For the critical use cases - you have planes. ✈️ For everything else - you have Freight. 🛳️

599

36 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

How India’s digital public infrastructure is enabling the emergence of state-backed ‘alt big tech’ "India Stack" Much of what was described as digital public infrastructure in this context flowed from India’s experiences with projects like Aadhaar for digital identification, the Unified Payments Interface for digital payments, and more recently, the account aggregator framework for data management in the financial sector. A host of such interventions built around the idea of standardisation of “open” application programming interfaces, or APIs, are collectively described as the “India Stack”. #digitalindia #upi #DPI #bigtech https://lnkd.in/g7tqyjSe

8

INDstocks by INDmoney saw almost 3x scale of concurrent users using the application through the last 3 days. As #India was getting prepared for its much awaited #generalElectionResults, the Engineering team at INDmoney embraced this challenge of getting ready for humungous scale. While many brokers had issues scaling for this traffic surge, primarily on June 3 (Monday), #INDstocks was quite resilient and gave all it's users highly reliable trading experience on all the three days (June 3 - 5). We have achieved this through a lot of best practices in #Engineering. All our Backends are independently scalable micro services, which gives us greater flexibility. We decoupled core components to a large extent, which allows seamless experience to our users on both Apps and website, even in the events of unprecedented scale. On days when volatility hits through the roof, an ultra-efficient square-off system which protects users from going into dangerous circuits and thereby losing money, is of paramount importance. Through our #INDprotect feature, our users' square-off orders were sent to exchanges in under 5 ms, which helped users reduce their losses significantly! At the market open, we saw close to 50% more price ticks from exchanges on June 4 and INDstocks users didn't have to face any latency in viewing the live prices on their web / mobile devices. Our #superFast #Pricer delivered all the ticks to users in under 1ms. #INDmoney #INDstocks #Election #Buzz #IndianEquities #Stock #Market #Economy Ashish Kashyap Nikhil Behl Mourya Venkat vasu choudhary Ashwin Sreekumar Sonu Singh Tejas Chhotani Deepak Gupta Kartikay Aggarwal Tushar Kumar Singh Deepak Pandey Nikhil Choudhary Sivajee Battina Pooja MD Harsh S. Aniket Pathak

218

8 Comments

Thrilled to support 💪 Perplexity return citations param - https://lnkd.in/g7kkztxX (+4 more updates 👇) ⚡️ Azure - return response headers ✅ Streaming - break out of infinite model loops 🎉 Instructor - fix completion_cost calc 👀 [PR] logging by API key - https://lnkd.in/gAG5YKQa

6

The more I work with #Elixir the more I come to appreciate the language and its #community. We at DAT AI Systems Pvt Ltd did a major migration from #Kafka to #AMQP without any downtime or client impact 💪. #Queues are the unsung heroes of #reliable, #scalable, and #distributed systems. With the AMQP protocol, an open standard widely adopted across industries, you can easily integrate #messaging into your stack without worrying about vendor lock-in. Why should you use queues in your architecture? - Decouple services for better scalability and resilience - Handle #asynchronous workloads efficiently - Ensure message #reliability and guaranteed delivery - Benefit from strong AMQP support across cloud and on-prem environments As businesses grow, embracing #open standards like AMQP ensures #flexibility, #portability, and long-term #stability. Ready to #future-proof your system? Check out my latest blog on the benefits of queues and AMQP. dev.to: https://lnkd.in/gDvce_nb Medium: https://lnkd.in/gVp9euNu #AMQP #MessageQueues #CloudArchitecture #Scalability #OpenStandards #EnterpriseTech #DistributedSystems

16

3 Comments

I was hot headed, typically obnoxious, and overtly arrogant when I started working at my first job. Chirag and Pravin dai will attest to that 😂.   I remember this one incident very clearly. Mahesh was my QA guy and we were building #MONyog, an agentless #MySQL monitoring tool. We missed a customer meeting. Just plain forgot. Obviously the boss' "blood (was) boiling" and his "head exploding" -- phrases we heard a lot those days! He rescheduled the meeting and stood over us while we took the call.   The problem the customer was facing was that every time MONyog started up his network usage shot up for the first few minutes and used up most of the available bandwidth. He showed us #Nagios graphs. Without thinking I asked him what the size of his error log was. He said it was around 1TB.   Every time MONyog booted up, it was reading 1TB of data using SFTP. How did I know? I had no idea. Boss looked at me, "How did you know?". All I said was: "I felt it". Sufficed to say that my boss' blood was no longer boiling over, and his ire was redirected at Sandhya who built the feature 🤣.   It's been over a decade since that incident, and I was reminded of it as I was reading Daniel Kahneman's #ThinkingFastAndSlow. His work, it shines a light on the dark recesses of my mind. It's incredible how the brain works, and we owe it to ourselves to take care of it, to nurture it. I had no idea, when I picked up the book, that Daniel would pass away just a day or two later. God rest his soul. Share your story. Show up.

23

9 Comments

OTN requires new multi modal user agent (like browser in case of internet) for discovery and executing transactions. Imagine prescription is uploaded by someone from tier5 and User Agent will understand , discover the business serving with lower price and make an order. Multi modal User Agent makes OTN truly accessible to billion people. Every transaction on earth should be powered by OTN (b2c, b2b, b2g, g2g). So every business application(sap, netsuite, etc..) should act as an agent. Onboarding businesses should be uber easy like enter type of business and your catalogue is ready. (this is possible only with universal Item registry). This will be true semantic web.

8

To all founders, CTOs, Vp Engg... if any company/individual approaches you with the following claims 1. We will build ML solution even if you don't have enough data {data synthetic will not work unless you have enough evidential data to create} 🛑 2. We have built LLMs or we will help you build LLMs 🤥 3. We will provide AI solution for very cheap 😥 4. We will improve efficiency/ROI/CR by 50% from day 1 using our AIML solutions Then please RUN 🏃‍♂️ 🏃‍♂️ 🏃‍♂️ .... All these are nothing but diet pill ads 💊 #CTO #CIO #VP #AIML #GENAI

25

4 Comments

Exactly-once semantics of Apache Kafka (part two) Transactions: atomic writes across multiple partitions We previously explained idempotent producers to help produce messages with no loss and no duplication. https://lnkd.in/dtPq-RE7 We need transactions on downstream processing to establish strong delivery guarantees. What's the problem? Applications potentially deliver inaccurate results when: 1. the app crashes after writing B but before marking A as consumed, the app reproceses A once resumes resulting in duplicate B 2. in a distributed setting, instances crash and are replaced with other members, reading same input and produce same output, these are called zombie instances How Kafka Transactions solve these? Transactions enable atomic writes across multiple partitions and topics, allowing a group of messages all being commited together or none. An example app consuming message A at offset X and writing message B as a result of processing such that B = F(A), and since offset commit is another write to an internal topic, now an atomic read-process-write means writing message B along with committing offset X as a single unit. How Kafka delivers exactly-once stream processing? 1. Kafka offers Streams API shipped with strong delivery guarantees. committing state, output messages and offsets atomically 2. Streams library is unfortunately only available for Java. using local state store backed by Kafka topics called changelogs for rebalancing and recovery 2. ksqlDB on other hand as an optimized database for stream processing enables executing sql-like statements 3. ksqlDB offers stateful aggregations, materialized views, joins and data transformations all backed by topics so again transactions 4. statements can be executed through exposed REST APIs, ksqlDB-cli or ksqlDB server Next we talk about zombie instances that are fenced off using transaciton IDs helping app survive restarts, and then two-phase-commit protocol by transaction coordinator.

18

2 Comments

Yajnadevam is a software engineer how has worked in crypto graphy and has used Shannon's information theory to decipher Indus scripts. interesting video to watch. will try to find his research paper to read and get more details.... #crytptogram #linguistics #information #iconography https://lnkd.in/gBBsXaG6

8

Good read

1

ok time for new vocabulary. my buddy Vinod Sankar used the term hallucinations in the context of some LLM discussions we were having in a previous post. I had never heard of the word used in the context of LLMs(yes i am not upto date on terminology) so i asked my fav LLM what it meant. Here is the response "In the context of large language models (LLMs) like GPT, hallucinations refer to instances when the model generates text or information that appears plausible but is factually incorrect, misleading, or entirely fabricated. These "hallucinations" occur due to the nature of LLMs, which predict text based on patterns learned from massive datasets rather than understanding facts or accessing a database in real-time." Its brilliant usage and apt terminology. AI models are slowly getting a human face.

4

13 Comments

I feel this debate about whether LLMs are intelligent and can actually reason or not is rather moot for most practical applications. Even if LLMs are just pattern matching systems, if the pattern matching is happening over a vast output of human reasoning, does it matter ? LLMs may not solve entirely new problems from first principles. But if they can solve such problems for everyone, that have been solved at least once by someone, that alone is a massive increase of intelligence available on tap to boost productivity.

1

Join us for a conversation with Peggy Rayzis on building a "Developer Intent led" GTM motion! We explore the key drivers that led to the wild success of Apollo Graph QL... followed by an AMA Link to register in comments. Feel free to share any questions

7

2 Comments

Our Kumo.AI blog post on Hybrid GNNs is out. https://lnkd.in/gpzr8TiF We have seen success with our Hybrid GNNs across several marketplace problems. Hybrid GNNs are a novel innovation at Kumo.AI because they model both repeated and explorative user interactions within a single GNN framework. The GNN automatically infers whether a user tends to do repeat purchases or tends to be more exploratory (ie buying new unseen items) and learns a repetition scalar per user. What we love about this approach is that it takes away from the standard recommendation systems architecture where you have multiple candidate generators - some for new unseen items, some for repeat behavior and then have to combine them in a final multi-objective optimization framework. Such systems which are a combination of multiple systems are hard to maintain, re-train and operationalize. On the Kaggle H&M challenge where most winning teams usually have approaches that require large amounts of feature engineering and the final system is a complex ensemble of approaches needing months to build, the Kumo.AI approach which is a 3 line predictive query that runs a Hybrid GNN under the hood is top 1%. Thanks Matthias Fey and Weihua Hu for the algorithms and innovation.

158

2 Comments

😇 I am starting a new series - "𝗗𝗲𝘃𝗼𝗽𝘀 & 𝗣𝗮𝗮𝗦 𝗣𝗹𝗮𝘁𝗳𝗼𝗿𝗺𝘀", a new post series where I will be evaluating and comparing various platforms on 𝗵𝗼𝘄 𝗲𝗮𝘀𝘆 𝗶𝘁 𝗶𝘀 𝘁𝗼 𝗱𝗲𝗽𝗹𝗼𝘆, 𝗺𝗮𝗻𝗮𝗴𝗲 𝗮𝗻𝗱 𝘀𝗰𝗮𝗹𝗲 our software apps. Products I will be talking about -- Heroku, AWS ECS, EKS, Fargate, App Runner, Render, Railway, Rancher, Fly.io, etc 𝗧𝗵𝗶𝗻𝗴𝘀 𝗜 𝘄𝗶𝗹𝗹 𝗯𝗲 𝗰𝗼𝗺𝗽𝗮𝗿𝗶𝗻𝗴 𝘁𝗵𝗲𝗺 𝗼𝗻 - 1. Ease of use and developer productivity. 2. Customisability, tweakness, support of various features. 3. Pricing and cost I need to pay for these features. 4. Creativity and modern innovation growth in these platforms.. If you want me to write on any other platforms than mentioned above, please comment and I'll add them in the series 😇 First post coming today evening, click the 🔔 icon on my profile to stay updated 😇 Edit - First post link - https://lnkd.in/gy7bWnK4 #devops #paas #cloud #deployments

83

5 Comments

One of the key capabilities that sets Memgraph apart from the competition is our deep path traversal algorithms. Come read our latest blog explaining all about it: https://lnkd.in/gBtWPJuZ

27

2 Comments

Speaking at #DES24, Pavan Nanjundaiah, Head – Tredence Studio, said today’s data engineer needs a beginner’s mindset to adapt to the growing change. “Today’s data engineers should expand left and right. Focus on your core foundational skills because they are not going to change. You don’t need to handcraft your code anymore.   Understand what is the business problem you’re trying to solve. My recommendation to data engineers is to shift left, which is the person who gave you the requirement and probably understands the business a little better. By doing this, you’re now zooming out of a core data engineer’s profile and understanding the business and the longer impact you can create.”   Read more here: https://lnkd.in/gKhtBwp6 #DataEngineers #Tredence #BeyondPossible

201

1 Comment