Subhro Kanti Bera

For this iteration, I decided to test out a new model trained on drum breaks, courtesy of fellow creative coder, Abhinay Khoparzi. The signal chain mirrors that of Fractured Neurons #1, but the training dataset posed an interesting challenge due to its periodicity. When I fed periodic material from the glitch machine into the trained model, the latency from the parallel processed signal became quite noticeable when layered with the dry glitch machine signal. Initially, I attempted to implement a dynamic logic to address the latency in real-time, but the results were uninspiring. In the end, Occam's razor held true. Applying a static delay to the dry signal turned out to be the most effective solution. I think this iteration has a unique sonic signature. Some of the sounds evoke timbres that feel both familiar and alien, and there’s something about the overall rhythmic structure that subtly hints at a glitchy tabla progression—but that might just be my perception and biases at play. Excited to keep exploring the sonic possibilities that arise as this project evolves! #neuralnetwork #machinelearning #proceduralaudio #generativemusic #creativecoding #maxmsp #flucoma #ircam

184

2 Comments

In 2014, India's average data consumption per user was just 0.3 GB/month at ₹269/GB. Today, thanks to Modi Govt's push and players like Jio, users consume 30.3 GB/month at ₹10-13/GB. Remember when we rationed data usage to stretch it for a month? Now, everyone's a legal expert, taxation guru, election analyst, military strategist and more! 😆 We've come a long way from just worrying about onion and tomato prices to debating complex tax issues like Indexation. Cheers to progress and thanks to Modi Govt for making us a more developed nation!

681

57 Comments

🔍 #ProductTalkwithMalthi - New Episode coming soon: "All About Data" with Sathish KS 🔍 In this episode, Sathish K S, CTO of Zeotap, shares his insights on building impactful data strategies for AI products. We covered: 🎯 Key Insights: 🖊️ Data Strategy for AI: How modern data platforms help businesses harness data effectively and the key elements that differentiate AI data strategies from traditional ones. 🖊️Growth through Data Platforms: How Data Platforms are driving growth and engagement while balancing data privacy. 🖊️Breaking Data Silos: Strategies to unify data across organizations, maintain high data quality, and keep data strategies adaptable. 🖊️Data Product Manager – Roles and responsibilities of a Data PM Sathish brought valuable perspectives on using data as a strategic asset, making this a must-listen for anyone navigating the data-driven landscape! Enjoy this snippet from our Rapid-Fire round 🚀 👉 Stay tuned to find the upcoming episode at  https://lnkd.in/ga7iYsrU #DataStrategy #AIProducts #ProductManagement #BusinessGrowth #Podcast Follow us at https://lnkd.in/g_GuEsjy

35

9 Comments

"The great thing about the Hype Cycle is that when we place something on the peak, where GenAI was last year, then there's always going to be a move down towards the trough” shares Gartner expert Afraz Jaffri. Find out what’s next for GenAI in our podcast 🎧 https://gtnr.it/4cJw3ww #GenerativeAI #GenAI #HypeCycle #ArtificialIntelligence #Podcast

7

🚀 Exciting Milestone for A&J’s OCR Model! 🚀 We’re thrilled to share that our in-house machine learning model for Gujarati OCR built by Dhiraj Gore and Siddhesh Lihe is outperforming Google Docs' OCR in key performance metrics for older datasets. These metrics reflect the power of niche training and non-general models that have the potential of beating bulkier ones when trained on clean, good quality and niche data. Here’s a quick comparison of the results: 🔹 Match Error Rate (MER): Our Model: 0.1673 Google's Model: 0.3239 🔹 Word Information Lost (WIL): Our Model: 0.2608 Google's Model: 0.5296 🔹 Word Information Preserved (WIP): Our Model: 0.7392 Google's Model: 0.4704 🔹 Character Error Rate (CER): Our Model: 0.0267 Google's Model: 0.0776 It has been a challenging journey so far with lots of obstacles but working on novel, cutting-edge research is our bread and butter. We are excited to continue developing and refining our model, as we strive to deliver even more powerful, real-world solutions to our clients. Credits to Aadit Malikayil for working on the foundational model, Abhishek Narayanan for pointing us in the right direction to improve our detection modelling, Dhiraj Gore, Siddhesh Lihe, who've worked tirelessly to make this possible and lastly, our client who gave us the project and helped us throughout the whole process. #machinelearning #AI #innovation #dataanalytics #financialanalytics #technology

26

5 Comments

In 2020, I finally found some time to reflect on emerging technologies and consider Postman’s role in this ever-evolving landscape. As I reviewed my notes from that period, the usual suspects appeared: #blockchain, #web3, #eventArchitecture, #streamingTech, #meshCommunication, #serverless, #edgeComputing, and more. (Yes, lots of hashtags!) Interestingly, #generativeAI was not at the forefront of my thoughts, despite being aware of #machineLearning, #ETL and the associated challenges developers face around these. Now, as generative AI takes centre stage, it reminds me of the early 2000s when mobile computing was rapidly gaining traction. Innovation thrives on #collaboration, with engineers and builders sharing knowledge across organisational boundaries, unblocking each other, and driving progress. At the cutting edge of new technology, open collaboration significantly outweighs the benefits of closed-door development, especially if you’re not part of the foundational layer of the business model for that technology. (Remember the path of #MongoDB anyone?) I’m thrilled to witness engineers coming together to build towards an #AI-accelerated, #APIFirst future. ❤️ Let’s keep at it Ayush Garg Rohit Agarwal.

16

On AI Appreciation day, I shared my views with The Economic Times on how Artificial Intelligence, grounded in contextually-relevant information, is reshaping business dynamics. AI's ability to uncover real-time insights and democratize complex data is revolutionizing decision-making across all levels of organizations. However, to harness AI's full potential responsibly, businesses must invest in robust data integration and governance frameworks. Read more here: https://lnkd.in/gFAUbtaN

22

As global economies embrace AI, it's clear that government support is necessary to accelerate AI adoption for the benefit of the public. India emerges as a critical player in the Global AI race, with the government driving the country's AI transformation. By fostering public trust in technology, India's innovative Digital Public Infrastructure (DPI) is a crucial aspect of AI adoption. Reflecting on my tenure as CTO at the City of Toronto, our Digital Infrastructure Strategic Framework (DISF) set digital guidelines that the City Council unanimously adopted. Approaches of this nature underscore the pivotal role of governments in shaping digital infrastructure to support the enablement of current and emerging technologies. As AI integrates further into our lives, collaborative government-private sector solutions are essential to tackle society's challenges. Private, Public, and Academic sectors should continuously collaborate to harness AI's power for the greater good. #AI #DigitalTransformation #digitalinfrastructure #digitalpolicy

26

👵 Did you know, Lenskart.com's AI & ML can forecast a store’s revenue with 95% accuracy prior to launching? 😲😲😲 Read my latest article about how AI is helping the retail industry ✒️ - https://lnkd.in/gehr53RA #artificalintelligence #machinelearning #retail #lenskart #consumer #supplychain #technology #transformation

6

During a recent short trip to India, I was struck by the sheer number of friends reaching out to discuss #AI and its impact on #jobs. What surprised me most was that these conversations were with individuals who aren't typically immersed in tech discussions. Their curiosity was sparked by AI creeping into everyday conversations, particularly in how it influences opportunities for the next generation. There was a common theme: Many had their kids studying in top universities in India, Europe or USA. In the past couple years, these students are facing challenges securing internships and jobs, partly due to companies leveraging AI to streamline operations. (Sure, the media has its share of setting up a narrative too). It is indeed stressing out many students and in turn, their parents. I feel for them both. Many had simple questions to understand the basics. Given my limited time in the trip, my local college friends hastily arranged a video call —to be conducted in plain & simple Marathi, the local language —to demystify what AI is, what it is not and what it means for the jobs of the future. We expected a dozen families to tune in. Turned out, over 50 families across multiple cities and continents joined! The response was overwhelming, with many more requesting access to the recording afterward. I found it was fun (though rather tricky) to boil the whole phenomenon down to simple explainers in #Marathi. Today, I'm sharing this video here. To my Silicon Valley friends: I know I oversimplified some concepts for this audience on purpose. My goal was to equip parents and students with a basic level understanding of AI and its implications on jobs. Some of this is my personal speculation too in answering their questions live. I hope someone you know finds this helpful. Your thoughts and feedback are welcome in the comments below. YouTube Link:

28

5 Comments

Mastering the art of impactful communication is key to making a difference, especially when leading a team that is in the driving seat to understand and guide others through a phase of unprecedented buzz and hype around data and AI. I just did my part and am glad that I have something to show for it. There is already so much to catch up on, but taking a step back and reflecting on what is truly required made me realize this. This is a course that I recommend to everyone who wants to understand the right framework of communication or validate the assumptions that have helped them shape their communication style. Thanks to INSEAD and Ian Woodward for an incredibly well-structured program. #Leadership #Communication #ContinuousLearning #INSEAD #ProfessionalGrowth

20

2 Comments

The hardest part of messaging is saying something the audience truly resonates with. Salad nails it. The latest episode of Do You Even Resonate features Salad (volunteered by Prashanth S Shankara 🥗). Salad is cloud for AI/ML, targeting data scientists. They lead with a differentiated value proposition: "Save up to 90% on compute cost", and they add credibility by being specific: "from $0.02/hr". Their ICPs instantly resonate with this as they know the pain of expensive GPUs. Inevitably, the question arises: " There's no free lunch, so how can these guys do it so cheap?" Again, Salad does a good job here by addressing it: - they use consumer-grade GPUs - 90% of the world’s compute resources sit idle 20-22 hrs/day, and they take advantage of this resource Boom! While they nailed the hardest part (teasing the promised land their ICPs want), they have room to improve. 3 key things to address: 1. Security concerns "The primary question is how security is managed across this distributed network." "Where are these consumer GPUs? Are they from gamers who join a "computesharing" network transiently? How could that possibly be secure?" 2. Reliability concerns "My fear is that the GPUs will not be reliably available to me, and that I will be part way through a model generation and then the GPU will vanish as the person who owns it needs it at that time." 3. Trustworthiness concerns as the website uses poor imagery and sloppy language "The duplicated text and some awkward wording is off-putting." "The word choices are awkward, and the graphics feel unprofessional." Those should be easy things to fix. Want to be featured on #doyouevenresonate? Comment with your ICP and URL.

63

21 Comments

The Parity Consulting annual product salary survey is a key resource for product leaders and product professionals to understand salary market trends and benchmarks. Please complete the survey to ensure that the data set is comprehensive and the insights valuable. Victoria Butt #productmanagement #prodmgmt #productsalaries #salarysurvey

9

1 Comment

Been mussing on "AGI" - a specter that is re-emerging given all the GenAI hype. Here is a note I shared with a few colleagues at Forrester on the topic in response to a well meaning but misinformed blog on the topic: "Thanks….yes we are writing about AGI, but trying to do is in a more pragmatic way…this guy’s post rambles. Plus, he conflates superintelligence with AGI, I think that is a big mistake. People also conflate AGI with machine consciousness, which is also another huge blunder that leads down pointless paths of speculation. A simple definition of machine intelligence as ‘a machine’s ability to achieve its goals given the tools at its disposal’ eliminates both. A machine doesn’t have to be conscious of what it is doing, or “superintelligent” by exceeding humans at everything to be AGI. It must be capable of having goals (either programmed, learned, or inferred) and using tools to achieve those goals. The more generally a machine can work towards many goals – book me a trip, buy me a home, send my kids to college,  optimize this supply chain – the closer to AGI it becomes. No super intelligence or consciousness needed…it’s just software doing what it is programmed to do, achieve broadly defined and somewhat ambiguous “goals”. This does raise all kinds of safety issues – how will such a machine deal with ambiguity? What if its goals involve making a tradeoff that could do some harm to achieve?  How do we ensure such machines consider the entire universe of human values and concerns, so they don’t game systems to achieve goals causing big issues? How do machines deal with other machines and humans in “multi-agent” scenarios. Etc. Etc. This is not simple but it’s also not necessary to assume super intelligence , human-like attributes, or consciousness…

2

1 Comment

Sharing my learning experience over the years. When it comes to predictive models, It is neither about the perfection nor about chasing the high accuracy. It is about iteratively creating value in terms of MVP - Minimum viable product or MMF - Minimum Marketable features that is compelling & time bound. It is about pursuing excellence iteratively for creating value & certainly not about pursuing perfection as there is none. Predictive accuracy also can simply be better than a random guess (desired in most use cases) that has the intended impact & not about the model accuracy that meets the industry benchmenarks that does not translate into business outcomes. ⁉️ Agree ⁉️ #predictivemodels #datascience #datasciencewithmanju #datasciencegalaxy

6

With the rise of AI, D&A leaders are actively trying to learn more about how Generative AI (GenAI) can benefit their organization. Understand the key AI challenges and get key actionable insights here. Gartner for IT | #GartnerDA #Data #Analytics

2