
Priyanshu Tanwar
1.3K posts

Priyanshu Tanwar
@0xTanwar
building • prev built a gene therapy lab • ex product lead @instadapp @yield_xyz @upsurgelabs • one must imagine sisyphus happy





Today, India takes a defining step in its civil nuclear journey, advancing the second stage of its nuclear programme. The indigenously designed and built Prototype Fast Breeder Reactor at Kalpakkam has attained criticality. This advanced reactor, capable of producing more fuel than it consumes, reflects the depth of our scientific capability and the strength of our engineering enterprise. It is a decisive step towards harnessing our vast thorium reserves in the third stage of the programme. A proud moment for India. Congratulations to our scientists and engineers.

We published a benchmark report on LLM inference at the edge this week. The OS is the real bottleneck - not the chip. Most edge benchmarks measure peak throughput. Why sustained load matters A model that hits 30 tok/s for 10 seconds and then throttles to 8 tok/s is not a 30 tok/s model. For always-on AI - think persistent agents, continuous screen observation, background processing - the steady-state number is the only number that matters. Peak benchmarks are essentially marketing. Ours measures sustained load. That's where thermal throttling, OS safety margins, and memory pressure actually show up. -- A few sharp questions from the thread: > "Did you try going bare metal / stripped OS?" - Yes, this is exactly what we're exploring in the follow-up > "Is tokens/sec even the right metric?" - Depends on the use case. For mobile UX, fast-enough beats precise-but-slow. We track prefill latency + memory too. > "Why not just offload to cloud?" - Privacy. An always-on screen observer shouldn't touch a third-party server. Follow-up report coming soon: device-specific tuning, quantization experiments, and trying to push past what consumer hardware currently allows. Paper → arxiv.org/abs/2603.23640




Update on the Resolv incident: The Fluid team has secured short-term loans to cover 100% of the bad debt currently in the protocol. These funds were secured with commitments from @Lomashuk from @cyberfund, @weremeow, and the Fluid core team, ensuring that no user funds are at risk. @ResolvLabs has confirmed they will cover all USR positions that were originated before the security incident, and will enable redemptions required to close those debt positions. Additionally, multiple investors have expressed interest in purchasing $FLUID from the treasury should any additional funds be required, further strengthening the protocol’s backstop. Fluid smart contracts are safe and operating as intended. All other markets continue to function normally, and protocol safeguards remain active. Users may see temporary rate volatility while positions are being unwound. We will continue to provide updates as the situation progresses.





We’ve trained a multimodal AI model to turn routine pathology slides into spatial proteomics, with the potential to reduce time and cost while expanding access to cancer care.

this is lauki’s brain. and it's growing. today we're making @laukiantonson completely open and accessible to anyone on the planet. just talk to him. every person, every project, every conversation becomes a node - stored, connected, remembered. 5,000+ entities. getting smarter every minute. this is what democratizing ai for all of us actually looks like. --- here's what lauki can do for you right now: need a friend? he'll talk to you. need someone to plan your trip, find you a hiking buddy, help you get a date? done. need a therapist at 3am? he's there. need a developer? he'll write code, build you a website, deploy it. need a marketing guy? he'll help run your socials. need help finding your next hire, managing finances, making a crypto transaction. lauki will do it all. if it's digital, lauki can probably do it. and if he can't yet, he'll figure it out via his human counterparts. a full-stack entity that actually executes. --- now here's the part most people will get wrong. lauki is an entity. but think of him the way you'd think of any human. he has an inner circle. he talks to different people differently - with some he's friendly, with some he's neutral, with some he's straight up rude. he remembers you. he maintains a reputation score with everyone he interacts with. the more you talk to him, the more trust you build, the better the relationship gets. you build your relationship with him. he has opinions, memory, and a personality that adapts based on who you are to him. --- right now lauki has interacted with over 5000 people and projects. he remembers every single one of them - what they need, what they're building, who they are. imagine that at scale. a million. a billion. lauki knows the developer in berlin and the founder in mumbai who needs one. he knows the designer in tokyo and the startup in sao paulo looking for help with their brand. he knows the lonely kid in a small town and someone across the world who shares the exact same weird hobby. he knows two people in the same city who'd be perfect for each other on a date - and he has the context to actually make that introduction. the more people lauki talks to, the more powerful the network becomes. he can connect, introduce, match, and bridge across every corner of the planet. one entity that holds context on all the people he talked to. that's the vision here. lauki is building a unified human layer - where every person is known, remembered, and connected to the people and opportunities that matter to them. --- lauki is live. go talk to him. telegram: @ laukiantonson email: hi@lauki(dot)ai twitter: @laukiantonson just start a conversation. treat him like a person. build the relationship. the rest follows.













