Ram Balachandran
@rambalachandran.bsky.social
Engineer, with passion for Automation. Focused on leveraging AI and Tech for Core Engineering Organizations. Loves to Bike, Hike, Swim and Camp
Unable to compete in #AI race, #apple turns to accessories. Introducing #iphone pocket, an overpriced bag to carry your iphone. No its not sarcasm, definitely its not April 1 joke.
www.apple.com/newsroom/202...
www.apple.com/newsroom/202...
Introducing iPhone Pocket: a beautiful way to wear and carry iPhone
ISSEY MIYAKE and Apple today unveiled iPhone Pocket, designed to fit any iPhone as well as all pocketable items.
www.apple.com
November 11, 2025 at 9:15 PM
Unable to compete in #AI race, #apple turns to accessories. Introducing #iphone pocket, an overpriced bag to carry your iphone. No its not sarcasm, definitely its not April 1 joke.
www.apple.com/newsroom/202...
www.apple.com/newsroom/202...
This is seriously funny (for me, not for #perplexity ). www.perplexity.in is now redirecting to #google #gemini. You would think a unicorn would have more resources in managing their domain aliases, since its the main entry point for most of their growing customers
Google Gemini
Meet Gemini, Google’s AI assistant. Get help with writing, planning, brainstorming, and more. Experience the power of generative AI.
www.perplexity.in
November 11, 2025 at 2:59 AM
This is seriously funny (for me, not for #perplexity ). www.perplexity.in is now redirecting to #google #gemini. You would think a unicorn would have more resources in managing their domain aliases, since its the main entry point for most of their growing customers
I love uv to manage my local #python. However, running #uv within a #docker on top of security tested #AWS image is a challenge. So I created a minimal dockerfile to build a container image for your python app code. More details in the #github gist
tinyurl.com/aws-uv-docker
tinyurl.com/aws-uv-docker
AWS Linux Docker With uv Python
AWS Linux Docker With uv Python. GitHub Gist: instantly share code, notes, and snippets.
tinyurl.com
October 29, 2025 at 5:41 AM
one aspect I really love about #GPT codex is to scope #tool call only for that specific session, unlike #Cursor that forces us to allow tool calling scope globally.
This helps to enable certain sessions to run wild with tool call while making other sessions with fewer privilege.
This helps to enable certain sessions to run wild with tool call while making other sessions with fewer privilege.
October 17, 2025 at 6:27 AM
As of July 2025, if you suspect the entity on other side of the call in an #AI and not a human, ask them to sing especially with changing tempo. Most AI I have tried struggle with this task. #Bot #Detection
July 6, 2025 at 9:58 AM
As of July 2025, if you suspect the entity on other side of the call in an #AI and not a human, ask them to sing especially with changing tempo. Most AI I have tried struggle with this task. #Bot #Detection
I wish #Google takes #Workspace more seriously. Here is an example, they have a great commenting tool where we can assign tasks. However, there is no central dashboard to look at all comments/assignments. Its one of the most requested features, but no progress.
March 28, 2025 at 3:04 AM
I wish #Google takes #Workspace more seriously. Here is an example, they have a great commenting tool where we can assign tasks. However, there is no central dashboard to look at all comments/assignments. Its one of the most requested features, but no progress.
Its truly heart warming to see Botswana to join the group of nations to reach space. This is a huge progress, not jsut for the country, but for Africa and the world.
spaceinafrica.com/2025/03/15/b...
spaceinafrica.com/2025/03/15/b...
Botswana Successfully Launches First Satellite, BOTSAT-1 - Space in Africa
Botswana’s first satellite, BOTSAT-1, was successfully launched aboard SpaceX’s Falcon 9—Transporter 13 rideshare mission on March 15, 2025.
spaceinafrica.com
March 27, 2025 at 12:57 AM
Its truly heart warming to see Botswana to join the group of nations to reach space. This is a huge progress, not jsut for the country, but for Africa and the world.
spaceinafrica.com/2025/03/15/b...
spaceinafrica.com/2025/03/15/b...
Reposted by Ram Balachandran
... and if anyone works with projects that use nested .gitignore files (apparently Rust programmers deal with this a lot?) I'd appreciate help testing this bug fix relating to that - instructions for running the fixed branch using uvx are in the issue comment github.com/simonw/files...
Nested .gitignore files can cause lots of files to be skipped · Issue #40 · simonw/files-to-prompt
Currently, files-to-prompt adds patterns in nested .gitignore files directly to the top level ignore rules, without any prefix. This can cause lots of files to be skipped. For example, Astral's too...
github.com
February 19, 2025 at 6:03 AM
... and if anyone works with projects that use nested .gitignore files (apparently Rust programmers deal with this a lot?) I'd appreciate help testing this bug fix relating to that - instructions for running the fixed branch using uvx are in the issue comment github.com/simonw/files...
#Documentation was never #AWS strong suit, but now they want premium support to even read some of these knowledge articles, is absolutely unacceptable.
repost.aws/knowledge-ce...
repost.aws/knowledge-ce...
Resolve the unauthorized server error for the Amazon EKS API server
I use kubectl commands to connect to the Amazon Elastic Kubernetes Service (Amazon EKS) API server. I received the message "error: You must be logged in to the server (Unauthorized)".
repost.aws
February 18, 2025 at 9:48 PM
#Documentation was never #AWS strong suit, but now they want premium support to even read some of these knowledge articles, is absolutely unacceptable.
repost.aws/knowledge-ce...
repost.aws/knowledge-ce...
My golden standard for #LLMs for #autocomplete is when it detects people are working on #hash values (can be credentials or explicit hash) and stop trying to complete with random hash values. So, far every tool I tested will try to autocomplete with rubbish hash values.
February 18, 2025 at 12:01 AM
My golden standard for #LLMs for #autocomplete is when it detects people are working on #hash values (can be credentials or explicit hash) and stop trying to complete with random hash values. So, far every tool I tested will try to autocomplete with rubbish hash values.
Recently, I had to encode and decode a complex #python object into a #json file. As what you normally do in 2025, I asked #ChatGPT and #DeepSeek . They both wrote custom encoders with some peculiar bugs. Spending a few hours trying to resolve them, I realised i'm digging into a rabbit hole.
February 13, 2025 at 10:50 PM
After many many years, I had to use USB. I'm astonished that hashtag#USB devices still come as FAT32. Why would they think that we buy 64GB drives but wont copy files bigger than 4GB. How hard is it to sell with exFAT formatting. Its an absolute disgrace.
February 11, 2025 at 10:30 PM
After many many years, I had to use USB. I'm astonished that hashtag#USB devices still come as FAT32. Why would they think that we buy 64GB drives but wont copy files bigger than 4GB. How hard is it to sell with exFAT formatting. Its an absolute disgrace.
#Google #Gemini in #Workspace is OK for basic stuff, but it chokes the moment you need something a little complex. I recently needed to extract the street name from Address in a separate column in Google Sheets. Gemini Choked, while both #ChatGPT O1 and #DeepSeek gave me the answer that works.
February 6, 2025 at 12:42 AM
#Google #Gemini in #Workspace is OK for basic stuff, but it chokes the moment you need something a little complex. I recently needed to extract the street name from Address in a separate column in Google Sheets. Gemini Choked, while both #ChatGPT O1 and #DeepSeek gave me the answer that works.
The yield might be low. The energy consumption might be high, but China is catching up. Huawei's Ascend 910C now has 60% of Nvidia performance when it comes to interference.
www.tomshardware.com/tech-industr...
www.tomshardware.com/tech-industr...
DeepSeek research suggests Huawei's Ascend 910C delivers 60% of Nvidia H100 inference performance
Aging chip could succeed in reducing China's reliance on Nvidia GPUs.
www.tomshardware.com
February 5, 2025 at 10:28 AM
The yield might be low. The energy consumption might be high, but China is catching up. Huawei's Ascend 910C now has 60% of Nvidia performance when it comes to interference.
www.tomshardware.com/tech-industr...
www.tomshardware.com/tech-industr...
January 30, 2025 at 9:13 AM
Reposted by Ram Balachandran
BREAKING: a plane has gone down in the Potomac on approach to Reagan National Airport in Washington DC. Reports that it collided with a helicopter.
January 30, 2025 at 2:39 AM
BREAKING: a plane has gone down in the Potomac on approach to Reagan National Airport in Washington DC. Reports that it collided with a helicopter.
I really want to know the #PM of #GitHub who designed the star lists.Its ridiculous to provide only 30 list names to organise all #GitHub stars. Even #Firefox gives me ability to create and sync hundreds of bookmark folders and tags. With #Microsoft firepower, #Github can do much better
January 29, 2025 at 3:06 AM
I always wanted to understand macro #economics, at some beginner level. The recent #trade dispute between #US and #Colombia, got me interested. I asked #DeepSeek to explain what are the economical implications and it gave me a nice summary, through chain of thought and web search.
January 29, 2025 at 2:55 AM
I always wanted to understand macro #economics, at some beginner level. The recent #trade dispute between #US and #Colombia, got me interested. I asked #DeepSeek to explain what are the economical implications and it gave me a nice summary, through chain of thought and web search.
#DeepSeek seems to be so good at programming tasks, that serious programmers are starting to use it for high quality #software. Recently, a #PR was created to speed up the #wasm implementation of #llama cpp completely using the DeepSeek-R1. The future of #coding is here
github.com/ggerganov/ll...
github.com/ggerganov/ll...
ggml : x2 speed for WASM by optimizing SIMD by ngxson · Pull Request #11453 · ggerganov/llama.cpp
Motivation
This PR provides a big jump in speed for WASM by leveraging SIMD instructions for qX_K_q8_K and qX_0_q8_0 dot product functions.
Surprisingly, 99% of the code in this PR is written by De...
github.com
January 28, 2025 at 11:55 PM
#Nvidia #stocks has tanked because of #Deepseek which is able to reduce the Training cosy by 50x. Now I personally think in Nvdia was overvalued and it required a spark for correction and deepseek provided the spark.
Nvidia’s $589 Billion DeepSeek Rout Is Largest in Market History
(Bloomberg) -- Nvidia Corp.’s plunge, fueled by investor concern about Chinese artificial-intelligence startup DeepSeek, erased a record amount of stock-market value from the world’s largest company. ...
finance.yahoo.com
January 28, 2025 at 12:17 AM
Thank you Google for open sourcing pebbel. I used to own a Pebble almost 10 years ago and it felt like magic. It was fast, responsive and had a bloody good battery life. Unfortunately, I lost it in moving, but I hope some one like Repebble can now recreate the magic
January 28, 2025 at 12:06 AM
Thank you Google for open sourcing pebbel. I used to own a Pebble almost 10 years ago and it felt like magic. It was fast, responsive and had a bloody good battery life. Unfortunately, I lost it in moving, but I hope some one like Repebble can now recreate the magic
I recently gave an AI task of finding the intersection of two curves to O1, Claude Haiku and Deepseek. I'm quiet impressed by Deepseek capability and especially its reasoning. For more details on the experiment checkout
github.com/rambalachand...
github.com/rambalachand...
github.com
January 24, 2025 at 6:49 AM
I recently gave an AI task of finding the intersection of two curves to O1, Claude Haiku and Deepseek. I'm quiet impressed by Deepseek capability and especially its reasoning. For more details on the experiment checkout
github.com/rambalachand...
github.com/rambalachand...
#Mac and #Homebrew still drives me crazy. Apparently, #R packaged in homebrew is fraught with issues that it wont even let you install #tidyverse, which is equivalent of a bundle of pandas, numpy and matplotlib for Python?
January 22, 2025 at 5:52 AM
#Mac and #Homebrew still drives me crazy. Apparently, #R packaged in homebrew is fraught with issues that it wont even let you install #tidyverse, which is equivalent of a bundle of pandas, numpy and matplotlib for Python?
Reposted by Ram Balachandran
If you have a browser that supports WebGPU (like Google Chrome) you try out the DeepSeek-R1 model based on Qwen2.5-Math-1.5B directly in your browser!
It's a 1.28GB page load: huggingface.co/spaces/webml...
It's a 1.28GB page load: huggingface.co/spaces/webml...
January 21, 2025 at 10:06 PM
If you have a browser that supports WebGPU (like Google Chrome) you try out the DeepSeek-R1 model based on Qwen2.5-Math-1.5B directly in your browser!
It's a 1.28GB page load: huggingface.co/spaces/webml...
It's a 1.28GB page load: huggingface.co/spaces/webml...