rushi-luhar.bsky.social
Software engineer, science fiction fan, girl dad, motorcycle addict, photography enthusiast, cricket tragic, long-suffering Gooner. 🇮🇳🇬🇧🇺🇸🏍️
37 posts
60 followers
156 following
Regular Contributor
Conversation Starter
comment in response to
post
Very useful post Laurie - thanks. I find they are useful at traditional NLP tasks too - entity recognition, sentiment analysis etc. I suppose it’s a type of “turn text into less text” use case..
comment in response to
post
Enshittification in action..
comment in response to
post
It’s also been an engine for significant growth and productivity (and significant downsides). I think AI (or LLMs to be precise) are powerful general purpose technologies which already have boring and beneficial uses. I understand your point of view but have a different opinion 🤷🏾♂️.
comment in response to
post
We are still in the Model T era of LLMs. They will improve and by a lot. It is a fair assessment to say that moaning about today’s energy consumption is not indicative of where things will be even a year from now. In any case, thanks for the discussion!
comment in response to
post
But you are comparing a Model T to a pick up truck.. I agree that the energy use of AI needs to be discussed and the hype around these models is ridiculous. But I am also frustrated at the reflexive dismissal of capabilities of LLMs - which, today, are going to be the worse they will ever be!
comment in response to
post
This post is like taking the MPG of a car from 1925 and assuming that it will be the same for all cars in all perpetuity. The inference costs for using LLMs are declining very fast. Even the training costs (see DeepSeek ) are going down. A conversation needs to be had about AI, but this is FUD.
comment in response to
post
It just requires an LLM or an offshore dev making 50k armed with GitHub copilot. I think this debate is a sham and a smokescreen to distract from the coming apocalypse of knowledge work jobs globally. Elon has spent and is raising billions building xAI. Immigration is a convenient distraction.
comment in response to
post
Love the lines of the fence, the road and the river. Beautiful.
comment in response to
post
Yeah - it’s basically cable tv at this point 🤷🏾♂️
comment in response to
post
Sports. I pay for YouTube TV so I can watch sports that I am interested in and are not available on other streaming services.
comment in response to
post
Last week, with encouragement from a friend, I set out to a lake near Vadodara with my dad’s kit. My preference probably will remain people and places but I enjoyed myself. There is something about standing around with a huge and heavy zoom lens waiting for the light and the birds to be just so..
comment in response to
post
I think it just pushed into to spiral of existential despair. Why ChatGPT, why?!!
comment in response to
post
As data volumes grow exponentially, efficient data movement becomes critical. Engineers who optimize for data locality and movement costs will build more scalable, efficient systems.
comment in response to
post
The push to “distributed everything” often results in surprise bills and shocking performance. Take a company I consulted with - they moved their analysis workloads from a relational database to a hot new data platform. Before they could realise any performance gains they were facing data costs.
comment in response to
post
As an engineer we are taught to obsess about computational complexity and look to compute as the bottleneck. But it’s data movement that tend to make life miserable. From memory cache misses to coordinating data between micro services to linking up data centers with fibre. It’s a fractal problem.
comment in response to
post
Have you come across JIRA? 🤔
www.loom.com/blog/loom-at...
comment in response to
post
The answer was yes, and I did.
comment in response to
post
It’s a rough transition back - but you will also be surprised at how efficient you can be despite chronic sleep deprived. Good luck!
comment in response to
post
And so falls apart another season. It was beautiful while it lasted Arsenal 😞