Hacker News: piterrro

New comment by piterrro in "Show HN: TERMy – A fast terminal assistant that does not use LLMs"

piterrro — Fri, 04 Sep 2026 19:52:14 +0000

You can get determinostic output (mostly) by setting the temperature to zero. Using couple of other tricks you can get close to 100% of determinism with LLMs.

New comment by piterrro in "Invisible Companies"

piterrro — Thu, 03 Sep 2026 19:07:22 +0000

If you own or work for an invisible company, upvote this comment

New comment by piterrro in "P99 0 ms* autocomplete for 240M domain names"

piterrro — Mon, 31 Aug 2026 08:10:55 +0000

I just typed a random sequence of the characters, long enough to be certain such domain doesnt exist. No only, the browser send an autocomplete request for every keystroke but for each request it returned a set of proposed domain names (which I'm 100% certain doesnt exist). At this point, how do I understand which results are legit and which are fake? Also, it would be nice to highlight the typed part in the result set so I can visually see what matches exactly.

New comment by piterrro in "SQLite as a Document Database (2020)"

piterrro — Sun, 30 Aug 2026 10:55:51 +0000

I do that in psql and it works really well. But your post got me thinking since I need to find a solution to store content of multiple documents, have a way to do FTS as well as vector similarity. I dont need that for all of the documents at once - I need to do it either for one document or at most couple of documents.

Now I'm thinking I could have a separate database file per "batch", store it in object storage and then download on demand and query it as I want. This way I'll not bloat my primary storage size as well as I dont need a special vector DB since sqlite vector search will be enough for up to 50k vectors.

New comment by piterrro in "Show HN: A lightweight, stateless database for agent memory"

piterrro — Fri, 28 Aug 2026 14:20:03 +0000

I get the closed source open binary approach - I would test it if I woild be in a need!

New comment by piterrro in "RAG Is Simpler Than You Think"

piterrro — Wed, 26 Aug 2026 15:39:10 +0000

RAG only makes sense if you have an LLM review the results, pick the most relevant ones and iterate further if there's a need running another query and repeating the process. Raw dump of vector search (even with reranking) is asking for troubles (or rather weird user questions like 'why this crap popped up in the results?')

New comment by piterrro in "Show HN: Huzzah – a novel approach to coding with AI"

piterrro — Thu, 20 Aug 2026 21:32:51 +0000

You could write that pseudocode as a prompt for the agent and get the same result. Use plan mode to understand what agent wants to do.

Am i missing anything?

New comment by piterrro in "PostgreSQL for Everything"

piterrro — Wed, 19 Aug 2026 14:21:26 +0000

true to that - currently using psql (in a single monolithic codebase) as: sql db, json db, vector store, logs store, full-text search, queue, message bus.

multiple processes connected to it.

New comment by piterrro in "Geolocating a random island using geometry and CUDA programming"

piterrro — Wed, 19 Aug 2026 12:48:29 +0000

really impressive, could that be the way to locate yourself without GPS? assuming we know more/less where we are

New comment by piterrro in "Cloudflare's AI Psychosis"

piterrro — Sat, 15 Aug 2026 15:15:10 +0000

Companies specialized in building DCs and AI labs specialized in niche models?

New comment by piterrro in "Cloudflare's AI Psychosis"

piterrro — Sat, 15 Aug 2026 15:01:12 +0000

The problem is, beside their size, they have not moat to compete in AI space. Their data centers are spread across the world in other companys DCs (btw do they own any DC actually from top to bottom?).

New comment by piterrro in "Don't classify, hallucinate"

piterrro — Fri, 14 Aug 2026 12:55:55 +0000

I would propose the following, query vector store for 10 closest categories based on a query, feed it to an LLM, in the prompt ask it to produce a single digit 0-9 representing the number of the most appropriate choice. Use plain text prompt, dont inflate token count with JSON. There you go, you just drastically reduced the output pricing.

Additionally you could experiment with a reranker instead of an LLM or after reranking take top-3 results and then feed to LLM as input in order to reduce input token costs.

New comment by piterrro in "Choose Boring Technology (2015)"

piterrro — Thu, 13 Aug 2026 19:55:29 +0000

https://grugbrain.dev/ Similar on this topic

New comment by piterrro in "Mistral OCR 4.1"

piterrro — Thu, 13 Aug 2026 19:47:36 +0000

You can even run it on your desk if you want, a single gtx 4090 is enough. It can be fully air gapped.

New comment by piterrro in "Mistral OCR 4.1"

piterrro — Thu, 13 Aug 2026 19:30:47 +0000

Accuracy can have different dimensions, depends on what you can tolerate and whether you can detect it to apply more powerful methods.

Imagine you have a cheap and 99% accurate ocr. The other 1% you can detect and apply more powerful (more accurate but slower and more expensive) ocr method. What would you use? At scale these things add up.

New comment by piterrro in "Mistral OCR 4.1"

piterrro — Thu, 13 Aug 2026 19:04:37 +0000

Most use cases dont need that kind of accuracy, just doesnt justify the 3-4usd range. I build for that exact case (tender documents, we’re processing north of 100k pages per day), it doesnt need to recognize scanned written text from 1930s, its usually pdf/docs/scanned printed pages. The accuracy is great, bounding boxes are must have for proper grounding for building answers by LLMs. Tesseract was too slow and not enough in some cases (for example tables or images which we also recognize and describe)

New comment by piterrro in "Mistral OCR 4.1"

piterrro — Thu, 13 Aug 2026 18:36:29 +0000

For anyone interested, I have an ocr pipeline running on rented GPUs, doing around 1000pages for 0.05-01 usd with around 0.8 seconds per page with full bounding boxes support for grounding.

If you’re interested you can find contact to me via this profile.

3.5 usd/1000 pages is just too expensive…

New comment by piterrro in "Don't be a meat proxy"

piterrro — Mon, 03 Aug 2026 08:09:18 +0000

In these situations I always use AI to respond

New comment by piterrro in "Why we write our own C and C++ inference engines"

piterrro — Mon, 03 Aug 2026 06:27:38 +0000

Could this vllm port be faster to install? Im starting gpu machine multiple times a day and it takes 5 minutes to set vllm up. If Inise this port that time is minimized?

New comment by piterrro in "CRM: An open-source, agentic-first CRM"

piterrro — Sun, 02 Aug 2026 08:59:58 +0000

Just cant believe AI slop is getting 1k stars on github in two days. Nothing really innovative about this project, nothing you cannot do yourself in two days with a coding agent.

I wrote it before, OSS is changing in a way that will basically make it useless.