Srinivasa Gandhi Sampath

#LSPPDay9, #60DaysOfLearning2024 #LearningWithLeapfrog https://lnkd.in/gBSRZHQJ Continuation to Day-08 Created a complete Crud Application. Today's Learnings: IO stream to handle image data, ModelAttribute and ModelMapper, more on Thymeleaf, Use of Dto

2

BT150's Ashwin Rangan on CIO evolution, technology curves and what it means for genAI deployments https://bit.ly/4h4wqFi @ICANNCIIO has been in the CxO game for three decades @ICANN, Rockwell Intl., @Walmart & @BankofAmerica, has seen his share of tech cycles and genAI is just the latest.

3

Why? While discussing system design for both trading apps and stock exchanges, I strongly emphasize that they are weak cases for Cloud. While Cloud is really good for scaling and availability, it comes with a tradeoff for Latency. While temporary 300 ms+ latency is OK for most of the apps, a trader can lose a lot of money in that time frame. I am still of the belief that a broker should run its app as close to the stock exchange as possible and latency should trade off for availability . If a person is not able to place order at the. first step, it's still better than seeing order in "Pending" state and then getting executed while stock price has gone down already. https://lnkd.in/d6N9Rp6G

3

I have posted about how ChatGPT makes mistakes on simple questions if they are phrased as puzzle. This is called template fixation where it may get the right answer, but logic is wrong. Link: https://lnkd.in/gfKEzRJ6 ChatGPT: https://lnkd.in/gbDVdQs8 #llms #genai #generatieveai #capgemini #llm #opensourceai #capgeminiindia #ai #artificialintelligence #software

15

In this #AISummit India session, we will explore how @Latentview Analytics is using their platform to address challenges in developing secure inference solutions posed by data security and privacy, and integrating with NVIDIA NIM inference microservices to streamline post-analysis tasks by understanding natural language and making data retrieval more accessible. Learn more. https://nvda.ws/4f8MaFC

6

Completing this course on ChatGPT Engineering Examples, inspired me to use on a specific use case. As part of Sankara School Leadership Academy (driven by Sankara Alumnus), we the Alumnus, are in receipt of data of "Building Tiny Grit" from the students. The challenge was that all students were posting data on completion every day, and we needed an approach of how we could consolidate data into a matrix on who all have completed BTG on what days. The rows were the students and the columns were the days on which they had completed their BTGs. Prepared a set of 4 prompts, and I had realized the information in just 5 minutes. I am pleasantly surprised at the approach of GenAI solving multi-level use case driven model so quickly and effectively. I recall 2 years back, I had created an application capable of "channelizing" Whatsapp data during the covid period to make sense of data, but this took me and my team of 4 people 1 month to create. Now, I accomplished all of that in just 4 prompts and 5 minutes. 792 hours of work completed in 5 minutes! A big thumbs up to the power of GenAI and those who constantly train the LLM! 👏

7

In this day and age of fast facts, it is always important to check your sources. I’m sure this will keep happening and the moral here is to take information shared with some balanced scepticism. "It seems the most accurate comparison would be against first-time test takers or to the extent that you think that the percentile should reflect GPT-4's performance as compared to an actual lawyer; then the most accurate comparison would be to those who pass the exam," study author Eric Martínez, a doctoral student at MIT's Department of Brain and Cognitive Sciences said. “But it turns out the artificial intelligence (AI) model only scored in the top 10% when compared with repeat test takers. When Martínez contrasted the model's performance more generally, the LLM scored in the 69th percentile of all test takers and in the 48th percentile of those taking the test for the first time.” In this instance as with many others in the past: - The method of how the analysis took place was not well articulated - What they determined as a success criteria was not clear and straightforward - Over simplification coupled by the overall hypecycle of excitement of LLM made us forget to ask the right questions - Glad there are people who are working hard to provide deeper analysis and we need more of this. Check out the full article: https://lnkd.in/gQ3YUp6r

2

Congratulations to our third annual vCon winning team: Suraj S., NAGASAI SAIKAM, Divya Sakhare, Akshata Salunkhe! The challenge? Implement redaction of personal information from a WAV file, and stick it in a vCon. Here's a deep dive into the hack, a decent introduction to the kinds of problems vCons can help you solve. BTW, looking for summer interns that know something about vCons? vCons enable the responsible management of conversations, including the detection of personal information, the redaction thereof, and even an audit trail of AI interactions for trust and transparency.

15

1 Comment

Mamba architectures seem to be all the rage now to reduce transformer complexity. This article offers a very nice introduction to the architecture and the math behind it: https://lnkd.in/gUXbUfy5 At its core, Mamba seems to be a discretized version of the continuous time Kalman filter with some additional tricks on top of it.

56

2 Comments

Advancements in Natural Language Processing: Enhancing Machine Understanding of Human Language in Conversational AI Systems https://lnkd.in/ge6khK-A Dr. Kavita D. Hanabaratti, Dr. Ashwini S Shivannavar, Dr. Sujit N. Deshpande, Dr. Rajesh V. Argiddi, Dr. RVS Praveen, Dr. Suhasini A. Itkar. This paper aims to review the recent developments in Natural Language Processing and their implications to the improvement of current comprehension in conversational AI interfaces. The comparison of four leading NLP models, namely BERT, GPT, T5 and XLNet is carried out systematically with respect to the major tasks of conversational interfaces including text generation, sentiment analysis and question answering. Based on the performance predicate that I used, it is clear that BERT has a 91% accuracy level. 5%, GPT 88. 2%, T5 89. 6%, and XLNet 90. 3%. Precision scores were precise to the second decimal place for BERT at 92. of reserves as 1%, while keeping GPT at 87%. 9%, T5 at 88. 5%, while last comes the averagely performing XLNet at 89. 7%. On the aspect of recall rates, BERT had a slightly better performance at 90.8%, GPT at 86. 5%, T5 at 87. ; 87% BERT, 88% RoBERTa, 89% XLNet. 2%. The recognized F1-scores were as well highest in BERT where it obtained 91. 4%, and XLNet at 89%. 5%, T5 at 88. Metcash Group Ltd at 17, Coles Group at 28, Woolworths at 10, Metcash Ltd at 6% and GPT at 87. 7%. This paper shows that BERT surpasses GPT-T5 in terms of accuracy and precision; however, GPT and T5 are better suited for text generation applications. These models contain theoretical and practical value in analyzing their advantages and drawbacks, which makes the base for choosing the suitable tools of NLP for definite uses and developed future conversational AI appliances. Keywords: Natural Language Processing, Conversational AI, BERT, GPT, Model Evaluation. International Journal of Communication Networks and Information Security (IJCNIS), 16(4), 193–204.

4

Planning to performance test Gen AI application but worried about cost overruns ??? Read on my article on medium about how you can control LLM costs while performance testing.

11

Can we have LLM routers? Unify has created a solution which abstracts the LLM being used - allowing us to use the best of breed. https://unify.ai/ An interesting concept to combine multiple LLMs #llms #genai #generatieveai #capgemini #llm #opensourceai #capgeminiindia #ai #artificialintelligence #software

6

1 Comment

🚨 Cyberism: A New Disease in 2009, A Growing Epidemic in 2024 🚨 Back in 2009, I wrote a blog post titled "Cyberism: Absolutely New Disease" where I delved into the emerging phenomenon of digital addiction. Fast forward to 2024, and this issue has only magnified, becoming a significant challenge for today's generation. Full Blog: https://lnkd.in/gxbqF-QQ 🔍 Then vs. Now: The Evolution of Cyberism - 2009: The early days of social media and smartphones. The concern was just beginning to surface. - 2024: A world where digital devices dominate every aspect of our lives, from work to social interactions. 🚨 Why It's More Dangerous Today - Increased Screen Time: On average, individuals spend over 7 hours a day on screens, impacting mental and physical health. - Mental Health Crisis: The rise in anxiety, depression, and other mental health issues correlates strongly with increased digital consumption. - Sleep Disruption: Excessive use of digital devices, especially before bedtime, is leading to widespread sleep disorders. - Social Isolation: Ironically, while we are more connected digitally, we are becoming more isolated in the real world. 📈 The Need for Action Understanding the trajectory from 2009 to today underscores the urgent need for awareness and proactive measures. It’s not just about managing screen time but about fostering healthier relationships with technology. 🛠️ Steps We Can Take - Digital Detoxes: Regular breaks from screens to reconnect with the real world. - Mindful Usage: Being conscious of how and why we use digital devices. - Education: Raising awareness about the risks of cyberism and promoting healthy digital habits. Let's acknowledge the impact of cyberism and work towards a balanced digital life. Reflect on how this once "new disease" is now a major part of our daily lives and consider the steps we can take to mitigate its effects. Read my full blog post from 2009: "Cyberism: Absolutely New Disease" on bitssmittal.blogspot.com. Read full blog here: https://lnkd.in/gxbqF-QQ #Cyberism #DigitalAddiction #MentalHealth #DigitalDetox #MindfulTech

8

1 Comment

Book Review : How do we scale the platform? How do we accomplish more with less? How do we cut down losses and improve efficacy? How do we accelerate speed to market? Irrespective of the platform/product, these are the challenges that every architect grapples with. When I was deeply focused on software enterprise architecture, Anand introduced me to this transformative book detailing how "Black & Decker" overcame near-collapse, outperformed offshore competitors, and rivalled giants like GE to reclaim industry leadership by investing in Platform Architecture. Concepts of scalability, standardization, and modularity are the same regardless of whether it's software or manufacturing. If you are a product /platform architect, this is a must-have book in your collections. When you complete the last page of this book, be assured you will be at least one inch more knowledgeable than what you began with 📚 What's on your must-read list for architectural excellence? #ArchitecturalWisdom #ScalabilityMatters #LeadershipLessons #ProductThinking #ContinuousLearning

25

7 Comments

"The AI Imperative: Essential Advances in Quality Engineering" 🔥 by Nagabhushan Ramappa   Writers often go through a cycle of drafting and discarding, refining their thoughts until they crystallize into words on the page. What if LLMs mimic this behavior ? Nag had some interesting content which i can relate to (not just) this ! 𝐖𝐡𝐚𝐭 𝐄𝐧𝐭𝐞𝐫𝐩𝐫𝐢𝐬𝐞 𝐄𝐱𝐩𝐞𝐜𝐭𝐬 𝐟𝐫𝐨𝐦 𝐏𝐫𝐨𝐥𝐢𝐟𝐞𝐫𝐚𝐭𝐢𝐨𝐧 𝐨𝐟 𝐓𝐨𝐨𝐥𝐬 Enterprises expect a surge in specialized SaaS productivity tools. They foresee a need for an automated, end-to-end product reliability center, ensuring that from development completion to end-user delivery, every step is autonomous and seamless.    𝑮𝒓𝒐𝒘𝒊𝒏𝒈 𝑺𝒐𝒇𝒕𝒘𝒂𝒓𝒆 𝑪𝒐𝒎𝒑𝒍𝒆𝒙𝒊𝒕𝒚  Post-COVID, companies have accelerated software development, introducing more features in one year than in the previous twelve. Managing this complexity is crucial, as errors/failures can lead to escalations.   𝘛𝘳𝘢𝘥𝘪𝘵𝘪𝘰𝘯𝘢𝘭 𝘘𝘈 𝘢𝘯𝘥 𝘮𝘢𝘯𝘶𝘢𝘭 𝘵𝘦𝘴𝘵𝘪𝘯𝘨 𝘢𝘳𝘦 𝘪𝘯𝘤𝘳𝘦𝘢𝘴𝘪𝘯𝘨𝘭𝘺 𝘪𝘯𝘢𝘥𝘦𝘲𝘶𝘢𝘵𝘦 𝘧𝘰𝘳 𝘵𝘩𝘪𝘴 𝘨𝘳𝘰𝘸𝘪𝘯𝘨 𝘤𝘩𝘢𝘭𝘭𝘦𝘯𝘨𝘦.   𝑹𝒐𝒍𝒆 𝒐𝒇 𝑮𝒆𝒏𝒆𝒓𝒂𝒕𝒊𝒗𝒆 𝑨𝑰 Generative AI can automate testing ( or any other tasks IMHO) and streamline workflows, enhancing efficiency and reducing errors in complex software development. This is where MAGS help us in stitching and unifying multiple workflows.   𝑴𝑨𝑮𝑺 - 𝑴𝒖𝒍𝒕𝒊-𝑨𝒈𝒆𝒏𝒕𝒊𝒄 𝑮𝒆𝒏𝒆𝒓𝒂𝒕𝒊𝒗𝒆 𝑺𝒚𝒔𝒕𝒆𝒎  MAGS utilizes multiple agents, each specialized in a specific task, working independently and sequentially. Let take an example of using GenAI In essay writing.   A non-agentic workflow produces an essay in a single effort - no backspace. Whereas, an agentic workflow iteratively refines the essay, with various agents conducting web research, evaluating content, and making improvements behind the scenes, ensuring a better final product.   Applying MAGS to complex software testing workflows, we can leveraging specialized agents to automate testing processes, ensuring thorough coverage and reliability in software development cycles.   𝗔𝗜 𝗶𝘀 𝗳𝗹𝗲𝘅𝗶𝗯𝗹𝗲 - 𝗶𝘁 𝗰𝗵𝗮𝗻𝗴𝗲𝘀 𝗮𝘀 𝗼𝘂𝗿 𝘄𝗼𝗿𝗸 𝗼𝗯𝗷𝗲𝗰𝘁𝗶𝘃𝗲𝘀 𝗰𝗵𝗮𝗻𝗴𝗲𝘀 Distributed intelligence systems are advancing and evolving from deterministic workflow tasks and freestyle chat to goal-based objectives. They are maturing from embedded AI and workflow-based approaches to case flow and self-organizing workflows.  As Nag explained these concepts, Amit Kumar Das balanced off with a Demo on the subject swiftly. I was amazed looking at the logs and trying to infer what was going on behind the scenes.   Nag - I loved the way the session went and how the demo were organized. Thanks for bringing this (new) awareness on AI stuff once again. Very informative session ! P.S. Have a look into the comments , for some questions Nag had for us ( i am yet to figure tis out even after a yr ;)) #TribeQonfReporter #TheTestTribe #TribeQonf #AI #saiprasadsabeson

37

2 Comments