Anant Narayanan

Here's what I get from it: - big tech companies are literally burning billions on "A.I." without any clear ROI plan - organic "A.I." adoption is abysmally low by their own employees because it simply doesn't work as promised, which is hypocritically embarassing for them, so now they're desperately trying to mandate its usage to prop up their reputational bubble a bit longer - CEOs still don't understand how their own sausage gets made and still don't care about producing quality products that will delight the true customers (the end users) - we have passed "peak A.I." (as all the big tech companies are starting to panic at the realization they have vastly overextended themselves by vastly overinvesting in "A.I.") and are now beginning to coast down the other side of the roller coaster track

5

Some of the most meaningful engineering work happens in unexpected places. Rivian’s haptic steering wheel is a great example of a this—a meaningful innovation we didn't realize we needed! But one use and it's obvious this is a leap forward in usability. Vidya Rajagopalan, Rivian's SVP of Electrical Hardware, and her team blended hardware and software to build an adaptive system that gives drivers control and tactile feedback without distraction. It’s elegant engineering and an experience that feels immediately intuitive. Follow along for more: https://lnkd.in/gJUq4xyW

924

12 Comments

Yet to see a firm talk seriously about outcome growth from AI. It is all “output, output”. Which is ironic given that for over a decade software has enjoyed compounding output gains through abstractions, while outcomes have stayed flat or even declined. Nobody seemed interested in measuring that. Individual output gains can actually reduce the productivity of the system as a whole. More output does not guarantee better outcomes. My mental model is this: if we are creating software at ten to twenty times the pace we did before, then the upstream and downstream feedback loops would also need to scale by the same factor to make sure we are building the right thing. That is not only unrealistic but undesirable. When everything speeds up, the noise to signal ratio explodes, and it becomes harder to know what actually moved the needle and what is worth iterating on.

66

23 Comments

Developing a large ecosystem with both open- and closed-source components comes with some interesting challenges. Back when we were bootstrapping CKEditor, my purist mindset pushed us to develop each of our 30+ packages in separate repositories. The goal was to sort of "force" higher modularity and more responsible architecture for every component. As you can probably guess 😅, it did not scale. Later, we consolidated most of the packages into two monorepos – one open source, one closed source – while keeping a few granular packages separate where it still made sense. This helped a lot, but it didn’t fully solve the problem. We were still working across multiple repositories and kept running into issues like dependency coordination and lockfiles. Recently, we took another step forward. We migrated to pnpm and redesigned our overall approach. In this post, Filip Sobol explains the new setup implemented by the CKEditor 5 Platform team and what we learned along the way: https://lnkd.in/eHS28qxe

23

4 Comments

ACTORS + AI: Director/Actor/Producer/Writer AI Chef 🧑‍🍳 Hans Obma visits the Cozy AI Kitchen 🎂 https://lnkd.in/gjiAvEfP to share his insights on how AI has transformed his creative processes. You may have seen Chef Hans in such 📺 hits as Better Call Saul, Wandavision, or Narcos — and yes, he's probably the most handsome AI 🧑‍🍳 chef we've had in the kitchen to date :-). Hans needed to learn a new language ... how did he do it? He got multimodal AI models to help him speak Welsh. Hans needed a way to scale how he communicated with his stakeholders ... how did he do it? He created an authentic-to-him way of scaling by creating an instant entourage of supporting agents. HT 🧑‍🍳 Ross Heise 🧑‍🍳 Matt Scholz and the power of the Wisconsin network for bringing such a storied talent to the Cozy AI Kitchen! --- This 🍰 episode: https://lnkd.in/gjiAvEfP All 50 episodes: https://lnkd.in/g6upvbGX

89

11 Comments

From Mid-Scale Hustle to Big-Scale Systems — My Leadership Journey When I worked at a mid-sized company — McAfee (formerly Intel Security), life as an engineering leader looked very different. I owned the entire tech stack end-to-end for the Real Protect SDK, which powered both enterprise and consumer security products. Decisions were quick. If I had an idea, we could experiment and roll it out within days. The visibility was immediate — the team saw the impact directly, and customers felt it instantly. But there were trade-offs: scale was limited, resources were stretched, and reliability sometimes gave way to speed. I wore multiple hats — tech lead, architect, and often the “on-call fireman.” Fast forward to today at Amazon. The canvas is massive — whether rolling out Prime membership features or building a multi-tenant portal to enable supply chain services for 1P & 3P sellers. Systems serve millions of users globally, reliability is non-negotiable, and playbooks ensure consistency across teams. I learn from some of the brightest minds, and every design decision forces me to think about scale, latency, and resilience. But here too, there are trade-offs: processes are heavier, decisions take longer, and ownership is often of just one slice of a much bigger puzzle. ⚡ What helped me move faster in this environment: Clarity of priorities → tie every discussion back to business + customer outcomes Cross-team relationships → build trust before you need alignment Document → Decide → Drive → cut endless meetings with written context & trade-offs Bias for small wins → ship incremental value instead of waiting for “big bang” launches Empowering the team → unblock quickly and let them own execution 💡 My takeaway: Mid-scale sharpened my ownership, versatility, and bias for action. Big-scale honed my systems thinking, discipline, and focus on reliability. The best leaders blend both — carrying startup agility into enterprise scale. 👉 If you’ve made the shift between mid and big companies, what was your biggest “surprise learning”? #EngineeringLeadership #TechLeadership #CareerGrowth #EngineeringManagement #ScalingTech #BigTech #StartUpCulture #LeadershipJourney

23

1 Comment

I recently found myself exploring about an interesting AI system design question: If Sanskrit and English had equal amounts of digital content available to train large language models (LLMs), would Sanskrit be the better language because of its well-defined grammar and logical structure? Sanskrit is often praised for its systematic grammar, explicit case markings, and structured sentence construction, where grammatical roles like subject and object are clearly encoded in word endings rather than inferred from word order. Intuitively, that feels like it should make reasoning easier for AI systems. However, modern LLMs are not rule-based systems that “apply grammar” the way humans learn it — they are statistical models that learn patterns from massive data. English, despite its ambiguities and looser structure, has simpler morphology and fewer surface variations per word, which aligns well with today’s Transformer-based architectures and tokenization methods. Sanskrit’s rich inflections and combinatorial forms, while logically elegant, can increase modeling complexity unless the architecture is designed to leverage that structure explicitly. This leads to a deeper realization: language structure alone doesn’t determine AI capability — architecture design and representation strategy matter just as much. It also raises a bigger question for the future of AI: as we move toward neuro-symbolic and reasoning-focused systems, could structurally explicit languages offer new advantages?

22

2 Comments

Waymo’s recent robotaxi incidents surface one of the most important and least discussed questions in AI product design: What should a system do when it doesn’t know what to do? Waymo’s answer is refreshingly honest, i.e. stop. Its vehicles are explicitly trained to halt when uncertainty rises (power outages, broken traffic lights, unexpected human behavior). This isn’t a model failure. It’s a product decision. When the safest action is inaction, the system chooses restraint over guesswork. At the level of one car, this is rational. At the level of ten cars, still sensible. At the level of hundreds, it becomes a system-wide failure. Waymo optimizes for local correctness, i.e. when unsure, do no harm. But without orchestration above the vehicle, local safety becomes global brittleness. Correct behavior in isolation can cascade into city-scale paralysis. The real lesson: building AI products isn’t about maximizing intelligence. It’s about designing systems that behave predictably when intelligence runs out and understanding how those behaviors compound at scale. Contrast this with China’s Apollo Go (Baidu) and Tesla and you will see three different strategies with different trade-offs. Baidu reduces ambiguity upstream with an “infrastructure-first” strategy. They’re not just building autonomous cars, they’re building V2X cities where roads, signals, and networks talk to vehicles. When things break, the city steps in. Fleets are centrally directed into predefined “safe harbor” states. The result is system-level resilience albeit at the cost of heavy coordination, regulation, and long-term capital alignment. Tesla takes the opposite bet. It assumes ambiguity will reach the vehicle and designs for it. No smart roads. Vision-only autonomy with end-to-end neural networks learning from millions of edge cases. Tesla isn’t optimizing for certainty in the moment, but for learning at scale over time. The product takeaway is clear: robust AI systems can’t rely on a single philosophy. The most resilient products layer their defenses: + Local safety behaviors that fail predictably + System-level coordination that prevents cascading failure + Continuous learning loops that improve with scale In AI, uncertainty isn’t the edge case, it’s the design constraint.

15

3 Comments

Building products and systems at Google is complex. But raising three still small kids is objectively harder. Production engineering system is full of redundancy. It is reliable and even has some Up Time guarantees. The system “family” is so fragile and not redundant! Like even though there are two adults but we have pretty different non-overlapping functions - like caregiving or home maintainance (pretty stereotypical.. haha) ? :))) So one needs a lot of **luck** to run the system “family” successfully, and **positive attitude** is a most underrated but most impactful secret ingredient! #WomenInTech #Motherhood #SystemDesign #HumanCentric

8

Aakash Gupta posted a really interesting set of learnings sitting down with a fintech executive from Robinhood. One key learning that resonated with me is: re: Legal, "Assume Good Intent - They're not there to make the product worse. Their job is to make sure that we are safely shipping the products we want to ship in accordance with the regulations." A partner in Fraud once said to me, "Henry, don't think of us as a gate. Think of us as your business partners trying to give you the best advice. Imagine if our role didn't exist, it would be your job to think of all of the angles that might attack your product." That was phenomenal advice that helped me re-calibrate my working partnerships with legal, risk, compliance, and other teams.

6

1 Comment