LLM Inference Is Just an In-Memory Database anuganti.com/articles/llm-in… #EnterpriseAI #DatabaseKernels #AIInfrastructure #AIEngineering #GPUs #LLMOps

Not sure whats happening at ⁦@AppleSupport⁩ …2 hrs for the update and almost 16hrs+ for reindexing…this is poor rollout

Models Are Ephemeral Compute: Building the Decoupled Enterprise AI Architecture linkedin.com/pulse/models-ep… via @LinkedIn #EnterpriseAI #AgenticAI #AIArchitecture #InferenceRouting #OpenModels #AIInfrastructure #AIStrategy #DigitalWork

@"Why Enterprise AI Needs an Economic Layer"linkedin.com/pulse/why-enter… #EnterpriseAI

Inference Firewall: The New Perimeter for the AI-First Enterprise linkedin.com/pulse/inference… on @LinkedIn #genAI #security #inference #firewall #waf #aisecurity

Just watched #Durandhar movie. Big letdown personally, maybe due to too much hype. The only plot I really loved was the concept of an Indian person sitting on the opposite side of most devastating events none of us can forget, to connect with the audience emotionally. #Raazi is still far superior.

Due to geopolitical tensions and airspace closures, my SFO->BLR #United/#Lufthansa was delayed, last minute united put me into airindia which is a direct flight to BLR.. #airindia this flight takes Vienna, Austria as stop-over for refueling and adds additional 2-3hrs extra fly time

Morning walk near reformA mexico city.. lovely angel of independence

what an incredible work from #TheGreatFlood team… amazing screen play, vfx and whole narration style.. highly recommend and must watch …

My latest article on Enterprise AI Architecture Reset from LinkedIn: The 2026–2028 Architecture Reset linkedin.com/pulse/enterpris… #EnterpriseAIZ #AI

Thanks @Tesla - new version of self-driving is almost ready for wide release. I've owned 5 Teslas in the last 10+ years, never had complaints, very happy owner. Couple of minor tweaks @elonmusk @SawyerMerritt @teslaownersSV: 1. Within 1-2 miles of an exit, can the car avoid switching left in any mode? Moving left and back is difficult when the exit lane is congested unless there's a clear gap. 2. When in the leftmost fast lane and a lane splits for an exit, FSD often gets confused, requires intervention, and drifts right into the exit lane, leaving the carpool lane. This needs retraining. 3. Wipers are still annoying and reduce visibility when passing under trees or sun/shade changes at speeds >70 mph, even in bright daylight. Sensor logic needs tweaking. Love the new version so far!

really bad to see how @X is killing the reading experience by introducing new embedded mode without allowing users to use their default browsers or apps for links? forget about lock-in experience; it might lead to more churn and a drop in engagement. - @nikitabier

Circular financing killed trust in "real" demand. Hard to know what's organic vs. inflated. #AI isn’t going anywhere. But the path forward will be painful, many will suffer along the way. Only companies with deep engineering discipline with innovation are safe to watch.

Every quarter, hyperscalers spend billions on GPUs. But no one reports the metric that matters, "how many tokens they produced with how little GPU". The next advantage won’t come from more hardware, it’s Smart Inferencing. #SmartInferencing #AIInfrastructure #AIEfficiency #GPUEfficiency #InferenceOptimization #TechLeadership linkedin.com/pulse/smart-inf…

The #iPhone17ProMax is a beast performance-wise, but that shiny new #AI needs a filter. It's a bit cluttered and intrusive. Great response times can't fix clunky software. Also, the build feels less premium, despite the new unibody. #Apple

Beyond BI: Why the Data Value Chain Defines the Future of Enterprise AI linkedin.com/pulse/beyond-bi… #AI #EnterpriseAI