commands that we support<\/a>\n, (add, commit, etc.)<\/p>"},{"title":"What are your priorities?","link":"https:\/\/suriya.cc\/essays\/priority\/","pubDate":"Mon, 12 Jan 2026 00:00:00 +0000","guid":"https:\/\/suriya.cc\/essays\/priority\/","description":"This is such a front loaded question, because depending on where you are in life or where this question is being pointed at (family, career, etc.) and the timeline given It is going to be wildly different.\nBut it is important to take time to stabilize the ever growing entropy.<\/p>"},{"title":"The future of media","link":"https:\/\/suriya.cc\/essays\/media_future\/","pubDate":"Sat, 03 Jan 2026 00:00:00 +0000","guid":"https:\/\/suriya.cc\/essays\/media_future\/","description":"
It is now possible to create content that is, a few order of magnitude more than the total time available to all humans currently alive. and I’m confident a major percentage of them can be tailored to be interesting for a person with specific psyche. My opinion is that human attention is going to become more and more valuable, yet it is going to get very difficult to focus on something for extended periods of time. What does this mean for the media ecosystem.<\/p>"},{"title":"qmd for hugo","link":"https:\/\/suriya.cc\/tech\/tinkering\/qmd\/","pubDate":"Mon, 29 Dec 2025 00:00:00 +0000","guid":"https:\/\/suriya.cc\/tech\/tinkering\/qmd\/","description":"
I really love just taking notes, run code and having it in my blog.<\/p>\n
Quarto is quite cool, I end up committing both the .qmd<\/code> and .md<\/code> files generated from running quarto render<\/code> locally. (haven’t gotten to cicd yet.)<\/p>"},{"title":"poor richard's almanack","link":"https:\/\/suriya.cc\/books\/poor_richard\/","pubDate":"Sun, 28 Dec 2025 00:00:00 +0000","guid":"https:\/\/suriya.cc\/books\/poor_richard\/","description":"I love poor Richard’s almanack specifically because the wisdom hidden within the words is malleable and every time I come back to read it I take something new based on my life experiences since then. This book has become part of this meditative 1 - 2 hours I spend every 6 months. and I can definitively say that this is a must read for everyone.<\/p>"},{"title":"my mother's recipe","link":"https:\/\/suriya.cc\/food\/vani\/","pubDate":"Thu, 25 Dec 2025 00:00:00 +0000","guid":"https:\/\/suriya.cc\/food\/vani\/","description":"
There is work underway to convert this into english. just haven’t gotten to it yet.<\/p>\n
Some of these dishes are so nostalgic even when I read the name.<\/p>"},{"title":"media processing in the cli","link":"https:\/\/suriya.cc\/tech\/tinkering\/image-processing\/","pubDate":"Fri, 12 Dec 2025 00:00:00 +0000","guid":"https:\/\/suriya.cc\/tech\/tinkering\/image-processing\/","description":"
convert RW2<\/code> format to jpg<\/code> with imagemagick. while maintaining quality.<\/p>\nfor<\/span> f in *.RW2; do<\/span>\n<\/span><\/span> magick "<\/span>$f"<\/span> -quality 85<\/span> "GooglePhotosBackup\/<\/span>${<\/span>f%.*}<\/span>.jpg"<\/span>\n<\/span><\/span> exiftool -tagsFromFile "<\/span>$f"<\/span> -all:all "GooglePhotosBackup\/<\/span>${<\/span>f%.*}<\/span>.jpg"<\/span> -overwrite_original\n<\/span><\/span>done<\/span>\n<\/span><\/span><\/code><\/pre><\/div>convert mp4 into good quality video.<\/p>"},{"title":"hail mary","link":"https:\/\/suriya.cc\/books\/hail_mary\/","pubDate":"Mon, 24 Nov 2025 00:00:00 +0000","guid":"https:\/\/suriya.cc\/books\/hail_mary\/","description":"
Overall I loved this book a lot. bought me a new passion for reading fiction.<\/p>\n
astrophage.<\/h3>\n
Astrophage (Greek for “star eater”), is an interstellar bacteria. sort of like the bacteriophage, but if it could somehow store neutrinos and travel at the speed of light. quite tacky, but leads to a very interesting storyline because they come to the solar system they start consuming the sun and sun starts losing its light. but on one star this does not happen.<\/p>"},{"title":"how big things get done","link":"https:\/\/suriya.cc\/books\/how_big_things_get_done\/","pubDate":"Sun, 16 Nov 2025 00:00:00 +0000","guid":"https:\/\/suriya.cc\/books\/how_big_things_get_done\/","description":"
the core principle: think slow, act fast<\/h4>\n\n- The vast majority of big projects fail\u2014they are over budget, over time, and under benefits.<\/li>\n
- The pattern for failure is “Think fast, act slow.”<\/strong> A rush to start, followed by endless delays, problems, and cost overruns.<\/li>\n
- The pattern for success is “Think slow, act fast.”<\/strong> A long, careful, iterative planning phase enables a quick, effective, and predictable delivery phase.<\/li>\n
- Planning is cheap.<\/strong> Delivery is expensive, complex, and risky. Do as much work as possible in planning.<\/li>\n
- The example of the house renovation in brooklyn which started with just the kitchen, which got delayed by 18 months and $800,000 was pretty telling. It is important to do small experiments before starting something huge. Plan things and define the facts ahead of time.<\/li>\n<\/ul>\n
on planning<\/h4>\n\n- Ask “Why?”<\/strong> before anything else. Plan from the right to the left<\/strong>. First, define the objective and the desired success. Only then should you work backward to figure out the best way to get there. Do not start with a solution.<\/li>\n
- Take the Outside View.<\/strong> Your project is not unique. It is “one of those.”<\/strong> Counter your biases by examining the performance of a broad reference class of similar past projects.<\/li>\n
- Use Reference-Class Forecasting.<\/strong> Anchor your budget and schedule in the actual data from comparable projects. This is the single most effective way to improve forecasting accuracy. It accounts for the “unknown unknowns” that are implicitly present in the data from past projects.<\/li>\n
- Plan like Pixar.<\/strong> Good planning is not a static, bureaucratic exercise. It is an active process of experimentation (experiri<\/em>). Use simulations, physical models, and digital twins. Build and test prototypes. Create multiple, cheap iterations to learn and discover problems before delivery begins.<\/li>\n
- Build with Lego.<\/strong> The key to successful, scalable projects is modularity<\/strong>. Identify the smallest, repeatable core component of your project. Master it, then repeat. This creates a steep, positive learning curve, driving costs down and speed up. Ask: “What’s your Lego?”<\/strong> Projects with high modularity (solar, wind) have thin-tailed risks. Bespoke, “one huge thing” projects (nuclear power, high-speed rail) have fat-tailed risks.<\/li>\n<\/ul>\n
on people & execution<\/h4>\n\n- Hire a Masterbuilder.<\/strong> Find a leader with deep, tacit knowledge and a proven track record of delivering similar projects successfully. Then hire their team. Experience is paramount.<\/li>\n
- Get the team right.<\/strong> A great team will fix a mediocre idea; a mediocre team will ruin a great idea.<\/li>\n
- Build a single, determined organism.<\/strong> Align all contracts and incentives toward the shared goal. Create a single team identity and a clear purpose. Foster psychological safety so bad news travels fast.<\/li>\n<\/ul>\n
on risk & mindset<\/h4>\n\n- Your biggest risk is you.<\/strong> Human psychology, with its inherent biases (optimism, uniqueness, confirmation bias), is the greatest threat to any project.<\/li>\n
- Ignorance is not your friend.<\/strong> The idea that underestimating challenges is beneficial because it tricks us into starting difficult projects is a fallacy supported by survivorship bias. Data shows the typical “leap in the dark” ends in failure.<\/li>\n
- Watch your downside.<\/strong> Risk can kill you; opportunity cannot. For projects with fat-tailed distributions<\/strong>, where extreme cost overruns are a real possibility, you must focus on black swan management<\/strong>: identify what causes tail risks in your reference class and mitigate them to “cut the tail.” You cannot simply budget for them.<\/li>\n<\/ul>"},{"title":"1d convolution","link":"https:\/\/suriya.cc\/tech\/algos\/cuda\/06_1d_convolution\/","pubDate":"Wed, 12 Nov 2025 00:00:00 +0000","guid":"https:\/\/suriya.cc\/tech\/algos\/cuda\/06_1d_convolution\/","description":"
#include<\/span> <cuda_runtime.h><\/span>\n<\/span><\/span><\/span>#include<\/span> <iostream><\/span>\n<\/span><\/span><\/span>#include<\/span> <stdio.h><\/span>\n<\/span><\/span><\/span><\/span>\n<\/span><\/span>__global__ void<\/span> convolution_1d_kernel<\/span>(const<\/span> float<\/span>*<\/span> input, const<\/span> float<\/span>*<\/span> kernel, float<\/span>*<\/span> output,\n<\/span><\/span> int<\/span> input_size, int<\/span> kernel_size) {\n<\/span><\/span>\n<\/span><\/span> int<\/span> i =<\/span> blockDim.x *<\/span> blockIdx.x +<\/span> threadIdx.x;\n<\/span><\/span> int<\/span> output_size =<\/span> input_size -<\/span> kernel_size +<\/span>1<\/span>;\n<\/span><\/span>\n<\/span><\/span> if<\/span> (i <<\/span> output_size ) {\n<\/span><\/span> float<\/span> sum =<\/span> 0.0f<\/span>;\n<\/span><\/span> for<\/span> ( int<\/span> j =<\/span> 0<\/span>; j<<\/span>kernel_size; j++<\/span>) {\n<\/span><\/span> sum +=<\/span> input[i+<\/span>j]*<\/span>kernel[j];\n<\/span><\/span> }\n<\/span><\/span>\n<\/span><\/span> output[i] =<\/span> sum;\n<\/span><\/span> }\n<\/span><\/span>\n<\/span><\/span>\n<\/span><\/span>}\n<\/span><\/span>\n<\/span><\/span>\/\/ input, kernel, output are device pointers (i.e. pointers to memory on the GPU)\n<\/span><\/span><\/span><\/span>extern<\/span> "C"<\/span> void<\/span> solve(const<\/span> float<\/span>*<\/span> input, const<\/span> float<\/span>*<\/span> kernel, float<\/span>*<\/span> output, int<\/span> input_size, int<\/span> kernel_size) {\n<\/span><\/span> int<\/span> output_size =<\/span> input_size -<\/span> kernel_size +<\/span> 1<\/span>;\n<\/span><\/span> int<\/span> threadsPerBlock =<\/span> 256<\/span>;\n<\/span><\/span> int<\/span> blocksPerGrid =<\/span> (output_size +<\/span> threadsPerBlock -<\/span> 1<\/span>) \/<\/span> threadsPerBlock;\n<\/span><\/span>\n<\/span><\/span> convolution_1d_kernel<<<<\/span>blocksPerGrid, threadsPerBlock>>><\/span>(input, kernel, output, input_size, kernel_size);\n<\/span><\/span> cudaDeviceSynchronize();\n<\/span><\/span>}\n<\/span><\/span><\/code><\/pre><\/div>"},{"title":"color inversion","link":"https:\/\/suriya.cc\/tech\/algos\/cuda\/04_color_inversion\/","pubDate":"Wed, 12 Nov 2025 00:00:00 +0000","guid":"https:\/\/suriya.cc\/tech\/algos\/cuda\/04_color_inversion\/","description":"#include<\/span> <cuda_runtime.h><\/span>\n<\/span><\/span><\/span><\/span>\n<\/span><\/span>__global__ void<\/span> invert_kernel<\/span>(unsigned<\/span> char<\/span>*<\/span> image, int<\/span> width, int<\/span> height) {\n<\/span><\/span>\n<\/span><\/span> int<\/span> i =<\/span> blockIdx.x *<\/span> blockDim.x +<\/span> threadIdx.x;\n<\/span><\/span>\n<\/span><\/span> image[i