siscia

5 ms·

31.
by siscia 10mo ago
I will be crucified by this, but I think you are doing it wrong. I would split it in 2 steps. First, just move it to svelte, maintain the same functionality and ideally wrap it into some tests. As mentioned you want something that can be us
32.
by siscia 10mo ago
I am toying with something similar. However my approach would be to use duckdb and S3 over lambda. Leaving many of the concerns to the infrastructure. Like basically no OOM. No need to manage servers.
33.
by siscia 11mo ago
With my partner we have been working to invert the overall model. She started grading conversation than the students have with LLMs. From the question that the students ask, it is obvious who knows the material and who is struggling. We do
34.
by siscia 1y ago
[flagged]
35.
by siscia 1y ago
Just to echo the point of MCP, they seem cool, but in my experience just using a CLI is orders of magnitude faster to write and to debug (I just run the CLI myself, put test in the code, etc...)
36.
by siscia 1y ago
This is absolutely neat.
37.
by siscia 1y ago
I find myself on both sides actually. I did have some great luck producing quite useful and impactful code. But also lost time chasing tiny changes.
38.
by siscia 1y ago
Do you really? Frontier models seems remarkably similar in performance. Yeah some nuances for sure, but the whole article could apply to every model.
39.
by siscia 1y ago
Also, do you have a better way to spend that money?
40.
by siscia 1y ago
In fairness, their design does not seem to be regional. With problems in one region bringing down another, apparently not unrelated, region. With this kind of architecture, this sort of problems is just bound to happen. During my time in AW
41.
by siscia 1y ago
As engineers we are (rightly) very scared of AI taking our job. If I were a lawyer I would be even more scared.
42.
by siscia 1y ago
What I see working extremely well, arguably in a setting where cost was not really an issue was a much simpler approach. Keep compacting files at some regular cadence `t` and keep a bucket policy to delete files older than `t+delta+buffer`.
43.
by siscia 1y ago
Anyone who has tried to write formal verification will tell you that there is a WIDE gap between thinking and writing the specs. Any tool that makes formal verification more accessible, should be welcome. I believe the valuable part is how
44.
by siscia 1y ago
This problem have been solved by LSP (language server protocol), all we need is a small server behind MCP that can communicate LSP information back to the LLM and get the LLM to use by adding to the prompt something like: "check your A
45.
by siscia 1y ago
I wonder if it would make financial sense to offer the "skeleton" of the car for sale and let design studios integrate all the pieces.
46.
by siscia 2y ago
Found funny that for something that is pretty much a commodity at this point, adoption seems to be the most important metrics. Yes, there are differences between the models, and yes some may work better. But picking the model at this point
47.
by siscia 2y ago
To be fair, beside technical documents, most of the link from Google are not worth reading. It is mostly SEO spam, written by LLMs anyway. At that point I prefer the summary from Google. I am not defending the model, but it is just the real
48.
by siscia 2y ago
What are the stats of the new system? Cost per giga? Latency curve on read? Latency curve on write? Support for concurrency? Max throughput? I think the article is somehow shallow, but S3 is a very deep and flexible system. Of course, someo
49.
by siscia 2y ago
Thanks! I am looking explicitly for people who want to "run ffmpeg" I don't have the skills, nor the capital, to build a solution for the whole market of people that want to mess with videos. Just for reference Mix raised 177
50.
by siscia 2y ago
Thanks! Why you think that an UI would make sense? An UI like what? Something to drive the user toward what it wants to do?
51.
by siscia 2y ago
The plan would it be to charge it at cost plus. Whatever the user is consuming plus a markup.
52.
by siscia 2y ago
Are ffmpeg user interested in a cloud base solution? You push the input files, the command, and fetch the output when done.
53.
by siscia 2y ago
Would you, or anyone else, be interested in ffmpeg in the cloud? Connect credit card, open a web UI, send the command, the files, and eventually get the output?
54.
by siscia 2y ago
I see few blind spots from the write up. 1. Traffic for a new version was loaded up too quickly. I usually lobby for releasing updates slowly. This alone would have prevented the issue. 1. Tasks cannot fail under load. Load Shedding should
55.
by siscia 2y ago
In my experience, coding with AI is much more mentally taxing than coding without. But it is much faster. When I use AI I need to continuously review, direct, and manage the AI. I go through every change and I agree with them, updates nits
56.
LLMs Are Power Tools for Software Builders(substack.com)
4 points by siscia 2y ago|1 comments
57.
by siscia 2y ago
I wrote down the same concept as a more structured substack. https://slowtechred.substack.com/publish/posts/detail/154138...
58.
by siscia 2y ago
I am very biased on this topic. I have started using AI coding assistant and I am not looking back. This comes from an engineer that KEEP telling the junior on his team to NOT use GenAI. The reality is that those tools are POWER TOOLS best
59.
by siscia 2y ago
Why? Whenever using a model to be more effective as a developer I don't particularly care if the model is open source or closed source. I would love to use open source models as well, but the convenience to just plug an API against som
60.
by siscia 2y ago
Commercial service have a context of 64k tokens, which I find quite limiting. https://api-docs.deepseek.com/quick_start/pricing Running it locally is quite a bit beyond the scope of being productive while coding with A

More ›