Dhawal Shah

While most of edtech is focused on classes 11 to 12 and competitive exams where the money is, we are building for the segment that will truly lay the foundation for India’s future: classes 1 to 5. If children at this stage become literate and confident, they can drive societal change and empowerment. In our Nipun app, we are interweaving AI systems, and we are already seeing teacher hours decrease significantly while their output has increased manifold. During a recent ground visit to a school, we noticed a peculiar challenge: younger students create a lot of background noise, which our AI system was not handling well. As a result, the first words of students’ responses during assessments were often cut off. To fix this, we added a simple 3 second gap before every assessment, giving students time to settle and ensuring their first words are captured. We are also building a noise reducer, leveraging open source libraries, to further minimize background noise and make the assessments more reliable. Vowels of the People Association #VOPA #V_School #Education #SocialChange #BuiltForImpact #NonProfit #EdTech #FLN #Nipun_Maharashtra cc: Deva Mhatre prafulla shashikant Pranav Pandey Venugopal Mankani Pratiksha Makeshwar Amit Devadhe Rohit Thakur Sachin Vilegaonkar Rutuja Jeve

38

2 Comments

Personal Experience on Using BharatGen's PARAM 1 – A Step Toward Truly Inclusive AI Over the past few days, I've had the opportunity to explore PARAM 1, BharatGen's Indic-scale bilingual foundation model and I have to say, it's one of the most refreshing AI experiences I've had in a while. What stood out to me? -> True bilingual understanding: I could interact naturally in both English and Hindi sometimes even blending the two. Unlike many global LLMs, Param 1 didn't flinch. It understood the context, the tone and the nuance. -> India-centric training data: You can feel the difference. The responses feel grounded, culturally relevant, and more aligned with everyday Indian realities. -> Open and accessible:As a developer and AI enthusiast, having free access to a model trained on 5 trillion+ tokens with 2.9 billion parameters - that's a big deal. Whether it’s education, health, governance, or content creation the applications are massive. For us Indians, this isn't just a model - it's a foundation for inclusive digital transformation in India. If you're building in the Indic AI space, I highly recommend checking it out. Explore the model here: https://lnkd.in/dJVcTBCF #AIforBharat #BharatGen #Param1 #MultilingualAI #DigitalIndia #OpenSourceAI #InclusiveTech #LLMs #HindiLLM #AIExperience #IndianAI

12

In Ayushi Sinha's "why AI radiology hasn't worked" post, one part that came up is how existing clinical data is extremely messy and distorted the documentation is optimized for billing, it's full of copy-paste notes that bloat everything, and there's lots of legal/hedging language You can still use it, but it requires way more work than you think because you need to understand what documentation is taken, how it gets stored, and what you need to do to fix it for the specific kind of analysis you want to do

12

3 Comments

🚀 Announcing the National Testbed Program for AI in School Assessments 🇮🇳 In collaboration with Centre for Educational Development (CED Foundation) we are excited to launch AIAE – AI for Assessment Excellence, a first-of-its-kind initiative to bring AI-powered grading and remediation into Indian classrooms. 📍Soft-launched at the School Leaders’ Conference in Delhi on June 29, this program invites Grade 10 schools to join a pioneering cohort of 15 institutions that will experience the power of Deepgrade AI – automating descriptive answer evaluation, identifying concept-level learning gaps, and supporting targeted remedial planning. 💡 Powered by Smartail and mentored by Centre for Educational Development (CED Foundation) experts, this initiative aligns with the spirit of #NEP 2020, IndiaAI, and global movements like the UK’s AI grading framework – helping schools personalize learning, save teacher time, and enhance outcomes. 🏆 A special highlight: 5 schools from the cohort will get the opportunity to showcase their AI implementation journey in Dubai at an international gathering of school leaders, innovators, and policymakers. 🔗 To express interest or get involved, drop a message or write to us at reachus@smartail.ai #AIinEducation #NationalTestbed #Deepgrade #CEDFoundation #Smartail #IndiaAI #EdTech #CBSE #SchoolLeaders #Remediation #AIGrading #DelhiLaunch #InnovationInAssessment #DubaiShowcase #EdTechIndia #AIforAssessment Trusity Rajani Nalla Swaminathan Ganesan Sheetal Srikanth Kannan Ganesan I Fizul Sheriff Ravi Perumal DR. PRIYADARSHI NAYAK Amaresh D. Deepali Bhagat Nishad Hussain BharathKumar Radhakrishnan HEC Paris - Innovation & Entrepreneurship Institute Rijesh Sreedharan Dr. Marc A. Bertrand Sijoy Thomas Sura Hussein

40

1 Comment

Just attended the Bharat Bodhan AI Summit, and the most interesting idea for me was the Bharat EduAI Stack: an open, interoperable “AI rails” (DPI) for education - so schools, states, universities, and edtechs can build on shared foundations instead of creating parallel systems. Why it matters: 1. Plug-and-play ecosystem: common building blocks like models & inference for faster shipping and easier integration. 2. Scale with trust: baked-in focus on responsible adoption for children for education use cases 3. End goal: move from scattered pilots to measurable learning impact at population scale via shared infra and partnerships. If this gets executed well, it could become the “UPI moment” for education AI -less hype, more delivery.

29

1 Comment

AI in EdTech needs to be safe, trustable and explainable.  Most of today’s AI models rely on the transformer architecture, trained on vast datasets, brilliant at generating fluent responses, but often opaque and prone to hallucination. I recently came across Gyan, a Boston based company, exploring a neuro-symbolic architecture that works very differently. ⭐ No heavy training required. Instead of expensive data pipelines and model fine-tuning, Gyan just ingests domain-specific terminology and knowledge (an ontology), and is ready to perform. ⭐ Outputs are built on understanding language structure and meaning, not guessing. That makes them fully explainable, traceable and hallucination free. ⭐ Keeps data private and secure. Because it’s not pretrained on external internet data, there are no IP risks, no biases and no privacy risks. ⭐ Runs efficiently on standard CPUs, without specialized hardware. That means lower cost, lower energy and potentially deployable on edge devices. This approach doesn’t diminish the value of regular LLMs, they are fantastic at absorbing and synthesizing knowledge from large datasets. But when accuracy, compliance, and explainability are critical (in industries like education, healthcare, law and finance), neuro-symbolic models like Gyan’s may be a smarter fit. By eliminating training overhead, enterprises save time and money, while gaining trustworthy, mission-ready AI that integrates in days not months. #AI #NeuroSymbolicAI #ExplainableAI #ResponsibleAI #FutureOfAI #EdTech #EnterpriseAI P.S. Gyan, I couldn not find your LinkedIn handle to tag you, would love to hear your thoughts if my interpretation above is correct.

6

1 Comment

An awesome interview, as part of SIGPLAN blog with Prof Ranjit Jhala, Professor at University of California, San Diego. The topics ranged from philosophical discussions, grad school experiences, how he developed an interest in model checking and PL research and finally down to a cool demo of refinement types in Rust. Absolutely loved the demo and surely would be playing around with Flux. Some of my rough notes from the discussion : - Model checking and Floyd Hoare logic : FH logic is symbolic model checking where u r using assertions to represent sets of states and manipulating these sets of states symbolically (preconditions, post conditions etc.) - How lazy abstraction evolved from a talk that was led to thoughts on how to reuse parts of a search space that has already been explored in previous iterations. Frugality of abstractions - don’t do stuff unless u really need it .. did I hear “Laziness by default is not my cup of tea ?” - How liquid types evolved organically as a fusion of constraint solving (that type systems do), model checking and lazy abstractions. - Thoughts on type systems in Haskell and Rust - Rust type system is extremely complex yet the compiler produces so good error messages. And there is a compelling value proposition - you do the type checking and since there is no GC, your code runs very fast. - The cool demo on Flux, a language-integrated verifier for Rust that leverages liquid types, a form of dependent type system. It was also cool to see CoPilot auto-completing flux annotations ;-). First runs rust type and borrow checker and then take the type information and then do liquid types style reasoning using SMT solvers .. Talks about the challenges of implementing refinement types in an imperative language, and here is flux. - Philosophy of Verse as a unification of functional and logic programming. Has a very very sophisticated type system - we expect to hear more in the near future. - Interesting discussion on whether CoPilot like LLMs will stifle new programming language design and research (as we don’t have many examples in the wild to train on) ... and a lot more. Well worth the one and a half hours .. Link to the interview: https://lnkd.in/g-ygZxcw

10

On the Arrived podcast, Ankush Sabharwal, the founder of BharatGPT and CoRover shared a vision that goes beyond tech benchmarks. For him, the real power of AI isn't just efficiency—it’s tackling the "unsolvable" societal problems we’ve lived with for decades. The Vision: - Impact over Imitation: Don’t just build the "next OpenAI"; solve a uniquely Indian challenge. - Quality of Life: Use AI to improve daily living for every citizen. - Bold Thinking: Daring to fix problems we once thought were permanent. We aren't just predicting the future; we’re building the tools to fix the present. Which "unsolvable" problem should AI tackle first? Full Podcast link in the comments. #AI #SocialImpact #ArrivedPodcast #Innovation #soverignAI #Soverigndata #LLM #AIImpactSummit #AIindia

47

1 Comment

ambaricloud: 108 recipes. 🥭 https://lnkd.in/gS6fuAtu Ravi Kiran Gurram Anil Koripelly Bhushan Lakhe Raja Purkayastha When this recipe e-book was first published on January 1, 2025, our goal was to give engineers, architects, and data practitioners a hands-on guide to Apache Iceberg. In the months since, the pace of change has been remarkable. Platforms such as Snowflake, Databricks, Dremio, RisingWave, ClickHouse, Cloudera, Lakekeeper, Trino Software Foundation, and Amazon Web Services (AWS) have delivered significant new integrations, and Apache Iceberg itself has reached version 1.10.0, introducing capabilities that reshape how practitioners interact with the format. To reflect this momentum, we are undertaking a ground-up rewrite. Our priorities include: ✅ Iceberg 1.10.0 alignment and deprecation cleanup – removing outdated recipes that no longer reflect best practices while upgrading all demos, examples, and exercises to incorporate the changes introduced in Iceberg 1.10.0. For example, Snowflake’s support for catalog-linked databases to external catalogs makes earlier object integration recipes obsolete. ✅ Ecosystem validation – testing and refining recipes across Databricks, Snowflake, Dremio, and AWS to ensure seamless interoperability with Iceberg 1.10.0. We expect this rewrite to take two to three months. Once complete, this edition will provide readers with a technically accurate, future-ready reference for Iceberg 1.10.0 and its surrounding ecosystems--Snowflake, Databricks, Dremio, RisingWave, ClickHouse, Cloudera, Lakekeeper, Trino Software Foundation, and Amazon Web Services (AWS) Please 🕺 suggest 🔪 comment 🥃 recommend ✅ Spark Variant ---Iceberg1.10.0 spports now. https://lnkd.in/gBmaNrPH

11

4 Comments

🚀 सहज जीवन: probably the first open-source, collaborative, public, and live Marathi book, built transparently on GitHub. 🚀 The first part is a translation of Leo Babauta's "The Effortless Life" in Marathi, and the second part is open for contributions. I’ve already added a couple of chapters there myself. The idea is simple: anyone with useful insights about leading a good life can join and contribute. Contributions are voluntary, no money involved. 💡 Eligibility criteria for contributors: - Fluency in reading and writing Marathi - Basic GitHub literacy (fork, commit, PR). LaTeX expertise is not required - And yes, you must have read almost all of PuLa. That’s non-negotiable 😉 You can either refine the translation of the pending chapters or add fresh chapters of your own. Just fork the repo, add your changes, and send a PR. You can also  your add your name as co-author, thus build your own version of the book into a PDF, print it for yourself, and even publish it for wider circulation. I’m planning to self-publish a slightly updated version via Notion Press, which will naturally come with some costs. Feel free to do the same, the PDF is freely available in the repo. If you want to add a chapter, just place it at the end and submit a PR (Pull Request). If you are not comfortable with LaTeX or GitHub, you can share me the content via Google-Docs or email, I will add that on your behalf. 👉 If you’re interested in this whole “effort,” drop a comment below or message me your Gmail ID. I’ll schedule a short online meet or create a chat group to align next steps. IMO, this is not just a book, it’s a movement: bringing open-source culture to Marathi literature. Imagine being part of a collective that publishes a free, evolving, community-owned book! 📎 Repo name: yogeshhk/SahajJeevan (... on GitHub). 📑 Attaching the partial draft PDF here for a quick taste. (It’s a very rough draft, just to give you the idea.) जय महाराष्ट्र ✨

110

12 Comments

AI will be part of the school curriculum from Class 3 onwards, starting this academic session (2026–27) 🤖📘 For primary students, the focus will be on interactive learning to introduce key AI concepts like pattern recognition and logic 🧠✨; while middle school students will explore real-world applications 🌍; and high school students will engage with AI innovation and emerging technologies 🚀 In the previous post, we saw how AI is making inroads in personalised learning 🎯. Countries around the world are incorporating AI in education and training children so they can thrive in a competitive AI-driven world 🌐⚡ To be sure, the current Indian education system — anchored in boards like CBSE, ICSE, and state curriculum — has been lagging and needs change. Rote memorisation has long been prioritised over innovation and the instillation of scientific thinking from an early age 📚➡️💡. We can see the disconnect as many new graduates are completely unprepared for AI-era jobs. I mean, look at the textbooks — they still talk about word processors and Microsoft XL! 😅 The revised curriculum may eventually bridge the skill gap, but we must remain vigilant as other countries adopt innovative strategies. We will see the impact down the road — though not all experiments have been successful. The cautionary tale is South Korea, which hurriedly launched AI-powered textbooks only to roll them back in four months ⚠️📉 (see carousel). An early start also means nothing if it can’t be sustained or if it doesn’t suit the level of understanding of students. A lot depends on how we implement the new curriculum and how much practical knowledge will be offered vs theoretical knowledge 🛠️📖. Application-based learning has become critical because we need to build muscles around context and problem framing 🧩. Solutions may become a commodity, but the ability to understand context and define the right problem will remain key 🔑. I guess ancient philosophers across civilisations would have had the best time now than ever 😄🏛️ And education alone isn’t enough. We also need to fund AI research 💰, upskill the workforce 🎓, and boost public-private partnerships 🤝 to enable AI projects to scale and develop our own AI models 🇮🇳 India boasts of a massive tech workforce 💪. Now, it needs to replicate the strategy to create an AI-ready one. It will require planning, leveraging resources for technical education, making use of our demographic advantage, and bringing reforms to address skill gaps. We have to “Adapt” — not just “Adopt” — if we want our knowledge economy to evolve and sustain the change AI is bringing 🔄✨ The AI race has just begun 🏁 — but sustained efforts could open up opportunities that help us sprint ahead of the competition 🏆⚡ #AI #ArtificialIntelligence #AIInEducation #NEP2020 #NationalEducationPolicy #GlobalAIEducation

29