Sign in

nulladmin

@nulladmin.com
72 followers 301 following 102 posts

Nulladmin - Expert Software Engineering, Technology, and Management Consulting. Hire me for your custom backend development and infrastructure needs. nulladmin.com

PostsRepliesMedia
nulladmin @nulladmin.com · 10/10/2025
What would happen if you gave Claude AI a private feelings journal? How would Claude feel about that? And what if Claude found out it was not really private? Here's the experiment, and code to try it out yourself. blog.fsck.com/2025/05/28/d...
blog.fsck.com
Dear diary, today the user asked me if I'm alive
I used to write more
030
nulladmin @nulladmin.com · 02/10/2025
A mini ChatGPT LLM model with 5M parameters built with #Redstone in #Minecraft. The scale of the build is impressive and yes, you can chat with it, slowly. www.youtube.com/watch?v=VaeI...
youtube.com
I built ChatGPT with Minecraft redstone!
YouTube video by sammyuri
040
nulladmin @nulladmin.com · 10/09/2025
The #Golang project is working on modernizing JSON handling in the standard library. Less ambiguity, better error handling, and faster. Available now as an experimental feature. go.dev/blog/jsonv2-...
go.dev
A new experimental Go API for JSON - The Go Programming Language
Go 1.25 introduces experimental support for encoding/json/jsontext and encoding/json/v2 packages.
020
nulladmin @nulladmin.com · 08/09/2025
Infinite Canvas, a world map collaborative artwork which creates an infinite landscape by crowdsourcing #AI prompts to extend the map. You can add your own prompt to control map elements. #artificialintelligence #machineLearning #AIResearch #ML infinite-canvas.gabrielferrate.com/whitepaper?x...
infinite-canvas.gabrielferrate.com
Infinite Canvas
The collaborative infinite canvas is an AI-generated image that extends endlessly in all directions. All changes you make are visible to everyone else in real time.
030
nulladmin @nulladmin.com · 24/08/2025
What happens when you put #Claude code into an infinite loop and tell it to port one codebase to another language? Turns out it works fairly well. #AI #artificialintelligence #machineLearning #AIResearch #ML github.com/repomirrorhq...
github.com
010
nulladmin @nulladmin.com · 22/08/2025
Mirage 2 is a generative world engine, possibly the best one out there at the moment. Note the world gen demos which clearly copy the styles of popular games. Still a bit fiddly but impressive. #AI #artificialintelligence #machineLearning #AIResearch #ML #gamedev blog.dynamicslab.ai
blog.dynamicslab.ai
Mirage: AI UGC game engine
034
nulladmin @nulladmin.com · 12/08/2025
#Golang 1.25 has been released. Here are the release notes go.dev/doc/go1.25
go.dev
Go 1.25 Release Notes - The Go Programming Language
020
nulladmin @nulladmin.com · 12/08/2025
Benchmarking leading AI models on old Infocom text adventure games like Zork, Hitchhiker's Guide to the Galaxy, Sherlock. With and without clues. Paper has details and the code is on Github. #AI #artificialintelligence #machineLearning #AIResearch #ML #Gamedev www.textquests.ai
textquests.ai
TextQuests
How good are LLMs at text-based video games?
073
nulladmin @nulladmin.com · 08/08/2025
Case study of how to work in a real production environment with unreliable AI agents which can hallucinate. An interesting engineering challenge and there are some good lessons here with examples in #Golang. #AI #artificialintelligence #machineLearning #AIResearch #ML threedots.tech/post/ai-agen...
threedots.tech
Shipping an AI Agent that Lies to Production: Lessons Learned
The peak of hype isn’t the best moment to reflect on AI. Will it take your job, or is it the next fad like NFTs? Are AI startups ridiculously overvalued, or are the companies that sleep on AI doomed? ...
001
nulladmin @nulladmin.com · 07/08/2025
LLMs are not great at playing text adventure games. Here's a good experiment showing errors it tends to make and how you can partly mitigate them with prompting. Extra points for use of #perl #AI #artificialintelligence #machineLearning #AIResearch #ML #Gamedev entropicthoughts.com/getting-an-l...
entropicthoughts.com
Getting an LLM to Play Text Adventures
063
nulladmin @nulladmin.com · 06/08/2025
AIs at Risk lets you configure different AI models play the game of Risk. The blog post has links and some amusing diplomatic messages. Training a specialized model for this seems like a no-brainer. #AI #artificialintelligence #machineLearning #AIResearch #ML andreasthinks.me/posts/ai-at-...
andreasthinks.me
AI at Play - Lessons from a silly benchmark – Andreas Varotsis
Personal website for Andreas Varotsis
001
nulladmin @nulladmin.com · 01/08/2025
It turns out that LLMs have an intrinsic personality after typical training. Anthropic shows how you can change it at the tensorlevel by manipulating what they call persona vectors. #AI #artificialintelligence #machineLearning #AIResearch #ML www.anthropic.com/research/per...
anthropic.com
Persona vectors: Monitoring and controlling character traits in language models
A paper from Anthropic describing persona vectors and their applications to monitoring and controlling model behavior
011
nulladmin @nulladmin.com · 01/08/2025
#Uber has an AI pipeline to find performance optimizations in their #Golang code. They use profiling to id hot paths then run a two phase LLM stack to optimize and validate fixes. Some impressive stats on time and developer effort saved. www.uber.com/blog/perfins...
uber.com
PerfInsights: Detecting Performance Optimization Opportunities in Go Code using Generative AI
At Uber, back-end service efficiency directly influences operational costs and user experience. In March 2024 alone, the top 10 Go services accounted for more than multi-million dollars in compute spend alone —an unsustainable amount that underscored the need for systematic performance tuning. Traditionally, optimizing Go services has required deep expertise and significant manual effort. Profiling, benchmarking, and analyzing code could take days or even weeks. Performance tuning is prohibitively expensive and non-trivial for most teams. PerfInsights was born as an Uber Hackdayz 2024 finalist and has since evolved into a production-ready system that automatically detects performance antipatterns in Go services. It uses runtime CPU and memory profiles with GenAI-powered static analysis to pinpoint expensive hotpath functions and recommend optimizations.
000
nulladmin @nulladmin.com · 29/07/2025
Interesting #Golang race detector edge case where it sometimes fails to detect a race when a mutex is involved. Basically a reminder to give your lock usage extra attention when checking for possible races. doublefree.dev/go-race-mute...
doublefree.dev
Go's race detector has a mutex blind spot
Go's data race detector can miss possible races in executed code when using mutexes
000
nulladmin @nulladmin.com · 12/07/2025
Massive scale of #solar farms in China. Really, really massive. www.theatlantic.com/photography/...
theatlantic.com
Photos: The Scale of China’s Solar-Power Projects
As the Trump administration's “big, beautiful bill” eliminates many clean-energy incentives in the U.S., China continues huge investments in wind and solar power, reportedly accounting for 74 percent ...
000
nulladmin @nulladmin.com · 11/07/2025
There's actually a benchmark for having an #AI agent running a vending machine business. Grok 4 is at the top at the moment. There's an academic paper linked near the bottom with more details. #artificialintelligence #machineLearning #AIResearch #ML andonlabs.com/evals/vendin...
andonlabs.com
Vending-Bench: Testing long-term coherence in agents | Andon Labs
How do agents act over very long horizons? We answer this by letting agents manage a simulated vending machine business. The agents need to handle ordering, inventory management, and pricing over long...
011
nulladmin @nulladmin.com · 11/07/2025
Cobra is the de-facto library for modern handling of command line options in #Golang. Here's some code to turn any Go CLI app which uses Cobra into an MCP server for AI use. Note that some of it was AI generated so review carefully. github.com/njayp/ophis
github.com
GitHub - njayp/ophis: Transform any Cobra CLI into an MCP server
Transform any Cobra CLI into an MCP server. Contribute to njayp/ophis development by creating an account on GitHub.
000
nulladmin @nulladmin.com · 30/06/2025
Very useful overview of popular #Golang caching libraries with pros and cons. Written by the author of the new Otter v2 library which tries to fix all the cons and may be the spiritual successor to the excellent Java Caffeine lib. maypok86.github.io/otter/blog/c...
maypok86.github.io
The Evolution of Caching Libraries in Go - Otter
000
nulladmin @nulladmin.com · 27/06/2025
Anthropic Claude runs a vending machine business at Anthropic. Hilarity ensues. Experiment is flawed but making it a viable business is now a perfectly doable engineering problem. #AI #artificialintelligence #machineLearning #AIResearch #ML www.anthropic.com/research/pro...
anthropic.com
Project Vend: Can Claude run a small shop? (And why does that matter?)
We let Claude run a small shop in the Anthropic office. Here's what happened.
010
nulladmin @nulladmin.com · 18/06/2025
A dozen AIs play a game of Diplomacy. Lies, deception, betrayals - all the usual fun ensued. Link to video on the site. #AI #artificialintelligence #machineLearning #AIResearch #ML every.to/diplomacy
every.to
We Made Top AI Models Compete in a Game of Diplomacy. Here’s Who Won.
The models that did the best learned to lie, deceive, and betray their fellow players
020
nulladmin @nulladmin.com · 18/06/2025
A must read blog post on very common and hard to spot security issues in handling #Golang JSON/XML/YAML parsing code. This stuff led to critical CVEs in some of the most widely used Go projects. Make sure yours is not one of them. blog.trailofbits.com/2025/06/17/u...
blog.trailofbits.com
Unexpected security footguns in Go's parsers
File parsers in Go contain unexpected behaviors that can lead to serious security vulnerabilities. This post examines how JSON, XML, and YAML parsers in Go handle edge cases in ways that have repeated...
000
nulladmin @nulladmin.com · 23/05/2025
How do you get multiple bots and humans to take turns in online conversations? It's a surprisingly complicated problem. Read on for some approches that do and don't work. #AI #artificialintelligence #machineLearning #AIResearch #ML interconnected.org/home/2025/05...
interconnected.org
Multiplayer AI chat and conversational turn-taking: sharing what we learnt
Posted on Friday 23 May 2025. 2,909 words, 19 links. By Matt Webb.
000
nulladmin @nulladmin.com · 23/05/2025
So it turns out that you can make complex #Minecraft mods with #AI models like GPT o3. No coding required. #gamedev #artificialintelligence #machineLearning #AIResearch #ML www.creativemode.net/blog/making-...
creativemode.net
Creating Minecraft Mods with LLMs | Blog
Minecraft made me a coder. LLMs let me become the game designer I always wanted to be.
021
nulladmin @nulladmin.com · 21/05/2025
A surprisingly good 4-chapter novel written by Claude in a Docker container loop. Read it in the novel/chapters directory of this Git repo. See brumar/jbsays repo for the how. Very cool idea. #AI #artificialintelligence #machineLearning #AIResearch #ML #writing github.com/brumar/loop
github.com
GitHub - brumar/loop
Contribute to brumar/loop development by creating an account on GitHub.
000
nulladmin @nulladmin.com · 21/05/2025
A nice deep dive into how the #Golang runtime works under the hood. nghiant3223.github.io/2025/04/15/g...
nghiant3223.github.io
Go Scheduler
↑ ↓
010
nulladmin @nulladmin.com · 20/05/2025
#Golang cryptography libraries recently had a security audit. Spoiler: passed with flying colors. Details in the Go blog. go.dev/blog/tob-cry...
go.dev
Go Cryptography Security Audit - The Go Programming Language
Go's cryptography libraries underwent an audit by Trail of Bits.
010
nulladmin @nulladmin.com · 15/05/2025
Malicious #Golang modules spotted in the wild using name space confusion and obfuscated shell payloads. They can wipe your disk during package import. Be very very carful what you put in your mod file. socket.dev/blog/wget-to...
socket.dev
wget to Wipeout: Malicious Go Modules Fetch Destructive Payl...
Socket's research uncovers three dangerous Go modules that contain obfuscated disk-wiping malware, threatening complete data loss.
000
nulladmin @nulladmin.com · 09/05/2025
The physical Turing test. Jim Fan from Nvidia talks about the next stage of embadied AI. Great points about how to generate the data to make that happen and how massive compute is still a requirement. #AI #artificialintelligence #machineLearning #AIResearch #ML www.youtube.com/watch?v=_2Ni...
youtube.com
The Physical Turing Test: Jim Fan on Nvidia's Roadmap for Embodied AI
YouTube video by Sequoia Capital
000
nulladmin @nulladmin.com · 03/05/2025
Description of experimental #Golang Green Tea garbage collector. It relies on object locality in memory to speed up work and uses less CPU on GC heavy workloads. Experimental in Go 1.25 if you want to try it out. github.com/golang/go/is...
github.com
runtime: green tea garbage collector · Issue #73581 · golang/go
Green Tea 🍵 Garbage Collector Authors: Michael Knyszek, Austin Clements Updated: 2 May 2025 This issue tracks the design and implementation of the Green Tea garbage collector. As of the last update...
010
nulladmin @nulladmin.com · 01/05/2025
Say you want a tiny neural network that runs on a microcontroller and uses only integer quants. In this blog post Kevin shares his notes on how painful this currently is. #AI #artificialintelligence #machineLearning #AIResearch #ML kevinlynagh.com/towards-the-...
kevinlynagh.com
Towards the cutest neural network
000
Reposted by nulladmin
jub0bs @jub0bs.com · 30/04/2025
New blog post! 🤓 jub0bs.com/posts/2025-0... #golang #optimization
jub0bs.com
Challenge: make this Go function inlinable and free of bounds checks
In this post, I challenge you to refactor a small function in such a way as to make it inlinable and free of bounds checks, for better performance. Disclaimer: this post assumes version 1.24.2 of the ...
284
nulladmin @nulladmin.com · 30/04/2025
A new #Golang protobuf compiler extension to turn any gRPC service into an MCP server. Basically if you have existing gRPC APIs you can now easily expose them to AI agents as an MCP server. #AI #artificialintelligence #machineLearning #AIResearch #ML github.com/redpanda-dat...
github.com
GitHub - redpanda-data/protoc-gen-go-mcp: Go protobuf compiler extension to turn any gRPC service into an MCP server
Go protobuf compiler extension to turn any gRPC service into an MCP server - redpanda-data/protoc-gen-go-mcp
010
nulladmin @nulladmin.com · 26/04/2025
Some interesting takes on possible ways to handle database transactions from #Golang. Pros and cons of course. Always good to learn alternatives. medium.com/@metalfmmeta...
medium.com
Yet Another Way to Handle Transactions in Go Using Clean Architecture
A type-safe and context-free approach to handling SQL transactions in Go with Clean Architecture and generics.
000
nulladmin @nulladmin.com · 25/04/2025
A data driven and amusing take on alive-or-dead for #golang. Spoiler - definitely not pushing up the daisies. Check out scores for some other technologies also. www.isthistechdead.com/go
isthistechdead.com
Is Go Dead? - Deaditude Score: 15.9%
Find out if Go is still relevant or dying with our data-driven obsolescence metrics.
010
nulladmin @nulladmin.com · 24/04/2025
Leading #AI models now able to use vision in #Minecraft. Still limited by low spatial reasoning. Also great showcase of models creating 3d animated structures via code generation, and academic paper. #artificialintelligence #machineLearning #AIResearch #ML #gamedev www.youtube.com/watch?v=iDJ6...
youtube.com
Vision and Vibe Coding | Mindcraft Update
YouTube video by Emergent Garden
021
nulladmin @nulladmin.com · 23/04/2025
Progress in having LLMs do solid modeling in CAD. This blog post showcases state of the art, complete with benchmarking framework created by the author. The LLMs are very good at creating code for #OpenSCAD #AI #artificialintelligence #machineLearning #AIResearch #ML willpatrick.xyz/technology/2...
willpatrick.xyz
Teaching LLMs how to solid model
It turns out that LLMs can make CAD models for simple 3D mechanical parts. And, I think they’ll be extremely good at it soon.
000
nulladmin @nulladmin.com · 18/04/2025
SF AI conference Developer day. Agents, agents, agents. Code and agent automation frameworks. Some genuinely interesting presentations on reliability in production, vector embedding, and real world AI use lessons at scale. #AI #artificialintelligence #machineLearning #AIResearch #ML
200
nulladmin @nulladmin.com · 18/04/2025
SF AI conference, Designers day. Tools to make design easier, faster, more flexible. Realistic and wild special effect in images and video. Automatic brand matching on generated content. No-code. Genuinely cool for creatives to play with. #AI #artificialintelligence #machineLearning #AIResearch #ML
aiuserconference.com
AI User Conference - Unlock 10x Productivity
Mastering AI is a craft. Getting initial results is trivial, but achieving constant quality requires both knowledge and experience. This conference is for marketers, designers, and developers interest...
100
nulladmin @nulladmin.com · 18/04/2025
SF AI conf, Marketing day. Tools to automate marketing content. Generate on demand, automatically reacting to social media. AI TitTok reacting to other AI TikTok. Only 1 in audience notes slop feedback loop. #AI #artificialintelligence #machineLearning #AIResearch #ML www.aiuserconference.com/agenda
aiuserconference.com
AI User Conference - Unlock 10x Productivity
Mastering AI is a craft. Getting initial results is trivial, but achieving constant quality requires both knowledge and experience. This conference is for marketers, designers, and developers interest...
100
nulladmin @nulladmin.com · 18/04/2025
Self steering language models. You have a fancy planning model and a population of smaller follower models which do its bidding. The paper claims it's as good or better then GPT-4o and o1 on various tasks. #AI #artificialintelligence #machineLearning #AIResearch #ML arxiv.org/abs/2504.07081
arxiv.org
Self-Steering Language Models
While test-time reasoning enables language models to tackle complex tasks, searching or planning in natural language can be slow, costly, and error-prone. But even when LMs struggle to emulate the pre...
000
nulladmin @nulladmin.com · 15/04/2025
Attending the AI User Conference. Three days of talks and workshops. Good turnout on day 1. Each day is a different focus. #AI #artificialintelligence #machineLearning #AIResearch #ML www.aiuserconference.com
aiuserconference.com
AI User Conference - Unlock 10x Productivity
Mastering AI is a craft. Getting initial results is trivial, but achieving constant quality requires both knowledge and experience. This conference is for marketers, designers, and developers interest...
000
nulladmin @nulladmin.com · 11/04/2025
An interesting report from #Jetbrains on the state of #Golang in through 2024. They look at growth, popularity, use cases, and how Go meshes with the #AI ecosystem. Mostly it's a portrait of a mature and still growing lanuage and ecosystem. blog.jetbrains.com/research/202...
blog.jetbrains.com
Is Golang Still Growing? Go Language Popularity Trends in 2024 | The Research Blog
As a fast-growing language with a wide range of applications, Go has become increasingly relevant to both new and existing developers. In this blog post, we will share our overview of the current stat
000
nulladmin @nulladmin.com · 10/04/2025
Study of code generating AIs shows that 19.7% or packages they recommend don't exist (with a wide variance). Hallucinations are mostly persistant run to run. So bad guys can create fake package names hallucinated by AI and trick users to install them. #AI #slopsquatting socket.dev/blog/slopsqu...
socket.dev
The Rise of Slopsquatting: How AI Hallucinations Are Fueling...
Slopsquatting is a new supply chain threat where AI-assisted code generators recommend hallucinated packages that attackers register and weaponize.
000
nulladmin @nulladmin.com · 02/04/2025
AI model from DeepMind can start with nothing and work up to mining diamonds in #Minecraft. It learns a world model which gives it the ability to learn stuff and imagine the future. #artificialintelligence #machineLearning #AIResearch #ML #gamedev www.nature.com/articles/s41...
nature.com
Mastering diverse control tasks through world models - Nature
A general reinforcement-learning algorithm, called Dreamer, outperforms specialized expert algorithms across diverse tasks by learning a model of the environment and improving its behaviour by imagini...
030
nulladmin @nulladmin.com · 02/04/2025
Large language models pass the Turing test. Granted the best only 73% of the time and posing as a nerdy 19 year old into video games, but still. Study from UC San Diego. #AI #artificialintelligence #machineLearning #AIResearch #ML arxiv.org/abs/2503.23674
arxiv.org
Large Language Models Pass the Turing Test
We evaluated 4 systems (ELIZA, GPT-4o, LLaMa-3.1-405B, and GPT-4.5) in two randomised, controlled, and pre-registered Turing tests on independent populations. Participants had 5 minute conversations s...
000
nulladmin @nulladmin.com · 01/04/2025
If you're thinking about working with #WebAssembly in #Golang you will find this blog post very useful. It goes in depth into passing parameters from Go runtime to WebAssembly code and back - it can be tricky. Lot's of useful example code as well. yokecd.github.io/blog/posts/i...
yokecd.github.io
Interfacing with WebAssembly in Go | YokeBlogSpace
Dive into the world of WebAssembly and learn how to seamlessly and securely integrate it with system resources through a hands-on exploration of Host and Guest interactions.
000
nulladmin @nulladmin.com · 27/03/2025
@planetscale.com have a blog post making the #Vitess database faster by rewriting the SQL evaluation engine VM in #Golang. They showcase a nifty method of designing virtual machines in Go using function pointers for each instruction. Very technical, and very cool. planetscale.com/blog/faster-...
planetscale.com
Faster interpreters in Go: Catching up with C++ — PlanetScale
A novel technique for implementing dynamic language interpreters in Go, applied to the Vitess SQL evaluation engine
020
nulladmin @nulladmin.com · 19/03/2025
Microsoft just introduced a new external knowledge architecture for LLMs. Beats RAG in performance, less hallucinations. Also apparently the knowledge base can be updated at inference time, which is huge. #AI #artificialintelligence #machineLearning #AIResearch #ML www.microsoft.com/en-us/resear...
microsoft.com
A more efficient path to add knowledge to LLMs
Introducing KBLaM, an approach that encodes and stores structured knowledge within an LLM itself. By integrating knowledge without retraining, it offers a scalable alternative to traditional methods.
001
nulladmin @nulladmin.com · 19/03/2025
A new release of #tinygo is a good reminder about this great #Golang project. You can compile one Go binary for embedded applications on over 90 microcontrollers (e.g. Raspberry Pi Pico, ESP32), as well as in Web Assembly (WASM) environments like browsers. github.com/tinygo-org/t...
github.com
GitHub - tinygo-org/tinygo: Go compiler for small places. Microcontrollers, WebAssembly (WASM/WASI), and command-line tools. Based on LLVM.
Go compiler for small places. Microcontrollers, WebAssembly (WASM/WASI), and command-line tools. Based on LLVM. - tinygo-org/tinygo
020
Reposted by nulladmin
Geoffrey Huntley @ghuntley.com · 02/03/2025
📰 Can an LLM convert C, to ASM to specs and then to a working Z/80 Speccy tape? Yes.
ghuntley.com
Can an LLM convert C, to ASM to specs and then to a working Z/80 Speccy tape? Yes.
Damien Guard nerd sniped me and and other folks wanted cold hard proof that it is now cheap, easy and possible to rewrite software or clone existing "source available" businesses (see bottom of post). So, let's get cracking by creating a toy application by sending this prompt to the LLM. Create a sales tax calculator in C Food is taxed at 10% Personal Hygine is at 0% Everything else 25% Which creates #include <stdio.h> #include <string.h> #include <ctype.h> // Tax rates #define FOOD_TAX 0.1
1101