tschillaci
tschillaci on Hacker News
Posted by
tschillaci on Friday July 17, 2026 @01:03PM
from the show-hn dept.
https://github.com/colomalabs/coloma
Hi HN,Thomas here, I'm a Machine Learning, and I've been deploying models in prod for the last 5 years and decided it was time to open-source some of my learnings!In Coloma you'll find my vLLM tuner, I use it to ensure models don't OOM at runtime, and are optimized for the metric that matters the most for that deployment: low per-user latency vs maximum system throughput.
To make it easier the...
Half my requests get retried because they fail, I've contributed to a ticket in June, with no fix yet.
I quit my job to work to create a story game powered by AI. We have a few agents that are in charge of memories, perception, and unlocking game achievements. If that sounds cool you can give it a try here:
https://k58.duya.io
Thank you so much, I appreciate it! I've actually quit my job to build games!
Posted by
tschillaci on Wednesday July 23, 2025 @11:28AM
from the show-hn dept.
https://k58.duya.io/
I've been working on a game for the last few weeks, it's called Kettle-Station K-58 and is about talking to Alex, a space courier resting at your remote asteroid stop.The goal is to build enough trust with Alex to get him to open up about his hidden backstory.Alex is powered by a few AI agents that handle memories, intimacy, and achievements. The achievements agent analyzes the conversation...
This is a really exciting project! I love how it makes each message feel more valuable, like modern-day letters, but more convenient.
I've just released a little story game where the goal is to build trust with the main character to uncover his backstory. And now I'm trying to figure out how to get the word out, which I have 0 experience doing!
https://k58.duya.io
You will find many agents that solved (e.g., finished, reached high score) atari games, but there is still so much more work to do in the field. I wrote my Master's thesis on how to learn from few interactions with the game, so that if the algorithm is ported to actual robots they don't need to walk and fall for centuries before learning behaviors. I think there is more research to do on higher levels of generalization: when you know how to play a few video games, you quickly understand how to play a new one intuitively, and I haven't seen thorough research on that.
My team’s building micro AI agents that perform very specific tasks really well. Right now, we see ourselves more as a consulting firm, developing custom agents while we search for the right vertical to specialize in. Eventually, we’d love to turn that into a SaaS product. We only have demos to show yet, but I’d be happy to chat more about it :)
Posted by
tschillaci on Tuesday February 25, 2025 @10:40AM
from the show-hn dept.
https://huggingface.co/spaces/thomas-schillaci/roast_this_website
I discovered Gemini is quite good at roasting websites so I put up a little demo on HF spaces if you want to try it out.
I quit my job last December to start an AI x AdTech startup which didn't work out: we discovered a similar piece of tech was about to get released by Google AdX. So now it's back to finding ideas, I'm sure there are niche problems that can be addressed using AI agents, one idea we have is developing an AI agent enabling content creators to better connect with their community.