Augustus Franklin Diraviam

2% of the mission's total ₹10,300 crore budget -while 44% has gone to compute. "Investments solely in compute are inherently transient- their value fades once consumed. On the other hand, investments in datasets build durable, reusable assets that continue to deliver value over time," said Chaitanya. Shared my thoughts on voice AI and datasets. Paywall article. https://lnkd.in/gbSRxmSQ

19

I recently interviewed Harshit Omar, Co-Founder & CTO of FluidCloud, and asked him three direct questions about what recent cloud outages really mean for enterprises. His answers cut through the noise — on:
• Why “multi-cloud” often isn’t truly resilient
• Where most enterprises are still exposed to risk
• What real cloud portability actually looks like in practice Great read!

34

1 Comment

A very nice compilation along with observations and recommendations on Creating products/solutions from Research. IIITH is one of the 6 institutions studied (others being IISc, IITD, IITB, BITS and CCAMP).

18

1 Comment

Vianai Systems, Inc.'s badass CTO, Navin Budhiraja, doing his thing. Insightful interview on the technological, professional and societal implications of AI and the need for enterprise grade AI technology to truly drive business performance.

2

Ahead of last week's India AI Impact Summit, one of the largest conferences, Vinod Khosla, respected VC and co-founder of Sun Microsystems, sparked a debate with his prediction that IT and BPO services will disappear in the next five years due to AI. The Indian IT services industry has already seen a $50 billion market capitalization loss in 2025 due to growth uncertainties. Recent AI plugin launches with prospects of automation raise questions about the necessity of traditional IT services, as Agentic workflows begin to replace conventional IT systems. Phil Fersht elaborated on this topic, referencing a blog by Vijay Vijayasankar, emphasizing that the transition will not be straightforward due to procurement inertia, buying behavior, and the continued need for human involvement. Moreover, it will be an evolution just like for all other industries, rather than a collapse. Phil says: Vinod is correct about the velocity of change, while Vijay highlights the constraints that exist. It is clear that IT and BPO companies must adapt to capture budgets shifting toward Agentic applications rather than relying on SaaS or traditional app development. While I agree with all above, I would like to add that securing AI budgets is not straightforward. A notable trend is that enterprises are increasingly inclined to conduct AI development in-house, driven by employee enthusiasm for building AI skills. AI app-dev is not mundane work to outsource, it is becoming a core and critical competency to develop inhouse. This shift towards in-house development means less reliance on outsourcing. Additionally, traditional IT and BPO service companies need to adjust their expectations regarding profit margins, as maintaining high profits of 20+ percentage is no longer feasible. One former CEO of an IT services giant reiterated that profit margins is the first priority of the IT services companies. This is a critical time for research and development, skill-building, and competitive sales and they become often a victim to aggressive expectations in profits. I discussed the above debate and more from the AI Impact Summit, the Gemini 3.1 Pro launch, and the rise of Personal Agents in the latest episode of The AI Files. Please check it out and share your perspectives! Jaime Basilico #TheAIFiles #GenAI 🎞️ YouTube Channel: https://lnkd.in/enUN4PDK 🎧 Apple Podcast: https://lnkd.in/eAxumKFw

45

7 Comments

“WE DON’T UNDERSTAND HOW OUR OWN AI WORKS.” —Anthropic CEO That admission is not a flaw; it’s a RESEARCH PROMPT at the nexus of silicon and self-inquiry. 🔹 REVERSE-ENGINEER COGNITION FIRST Vipassana provides direct, time-stamped introspective access to phenomenological data. Its stepwise ATTENTION MECHANISM isolates breath, emotion, and memory the way a transformer weighs embeddings. 🔹 INSIGHT, NOT COLLAPSE Deep practice yields REPRESENTATIONAL TRANSPARENCY: cognitive subsystems keep firing yet are observed as constructs. That mirrors the goal of AI interpretability dashboards—no “model collapse” required. 🔹 ETHICS AS OPTIMIZATION In the brain, CETANA (intention) drives karmic feedback. In code, GRADIENT DESCENT updates weights. Same loop: lower loss, less dukkha. 🔹 CROSS-DISCIPLINARY INTERPRETABILITY Put meditators, neuroscientists, and ML engineers at one table. Shared metrics, shared mispredictions, better causal insight. 👉 NEXT STEP I’ve distilled these bridges into GenerativeMindOps: Self as a Generative Process. Grab below and/or DM me to join the peer-review circle where we stress-test these techno-spiritual hypotheses against live model logs. Let’s make the MIND its own unit test—and push AI interpretability forward together.

12

2 Comments

https://lnkd.in/gqJDGhv6 "In the realm of higher edu­ca­tion, India has a lot to learn from the U.S., espe­cially in terms of pro­mot­ing lib­eral edu­ca­tion, crit­ical think­ing, inter­dis­cip­lin­ar­ity, indi­vidu­al­ised learn­ing, and hol­istic and trans­form­at­ive edu­ca­tion" Prof. V Ramgopal Rao

1

Microsoft, Google, AWS, among others are investing big in India, a move that could support AI infrastructure developments and meet local data hosting requirements. Just last month, Google announced $15 billion, its single largest investment in the country . Earlier, AWS pledged $12.7 billion in cloud infrastructure by 2030, while Microsoft unveiled a $3 billion plan to expand its AI and data center footprint over the next two years. The story takes a look at global big tech's big bets in India:

38

2 Comments

I’ve released a short analysis of the Puloma account from the Markandeya Purana, focusing on decision-making under pressure and the ethics of speech. The case shows how an accurate statement can still cause severe harm when context, consent, and timing are ignored. I examine the roles of Agni, Bhrigu, and Puloma to outline practical principles for truth-telling, accountability, and conflict resolution. https://lnkd.in/g5xws-BS

1

AI's New Power Brokers: India's Historic Delhi Pact, Harvard's ML Systems Bible, and Claude in Your Pocket The AI world just got reshaped—literally. GitHub trending (Feb 19, 2026) and breaking news from India's AI Summit show three seismic shifts: geopolitics bending Big Tech, education leveling up, and your terminal going mobile. No fluff, just moves that matter. Lead: The New Delhi Frontier AI Impact Commitments just dropped. Google, OpenAI, Meta, Anthropic, Microsoft plus Indian players (Sarvam, Gnani, Bharatgen) signed PM Modi's voluntary pact at the India-AI Impact Summit. Focus: real-world AI usage analysis, multilingual eval for Global South contexts, democratizing AI away from "few hands." Pichai, Altman, Amodei all present—tensions visible (no handshakes for cameras). This is Big Tech bowing to emerging market pressure, opening R&D doors. Investors: India TAM just became unavoidable—$1.3B+ AI revenue forecasts in play. Source: Indian Express (2026-02-19) - https://lnkd.in/gV87NUgp Harvard just dropped the ultimate ML systems bible. harvard-edge/cs249r_book: "Introduction to Machine Learning Systems"—open-source, MIT Press publishing 2026. Goes beyond TinyML: training at scale, inference optimization, data pipelines, ML ops. 20+ chapters from Prof. Vijay Janapa Reddi's course, community-driven. Trending because every engineering team needs this—foundational for building production AI. Builders: clone, learn to ship. Operators: onboard faster. Investors: This is the textbook every ML engineer will cite—moat in education. Source: GitHub Trending Daily (2026-02-19) - https://lnkd.in/gDuWXncr RichardAtCT/claude-code-telegram: Run Claude Code from your phone via Telegram. Full AI assistant + session persistence—no laptop needed. Prompt from anywhere: review code, debug, write PRs. Hot for remote devs, on-call engineers. pip install, connect bot, terminal in your pocket. Pro: Multi-session memory. Builders: extend for team bots. Operators: emergency fixes from beach. Investors: Mobile AI agent play—$5B+ market. Source: GitHub Trending Daily (2026-02-19) - https://lnkd.in/g4EXiPC9 2026 isn't about bigger models—it's about who controls the stack. India just claimed a seat. Harvard's teaching the next generation. Your phone just became your dev environment.

3