Blue_Cosma

6 ms·

1.
by Blue_Cosma 8mo ago
yes, private AI is not financially viable right now vs. APIs. The main reason to invest in such hardware is privacy and security.
2.
How many users can a $50K AI workstation serve? Benchmark data(old.reddit.com)
2 points by Blue_Cosma 8mo ago|2 comments
3.
by Blue_Cosma 9mo ago
Thanks a lot for sharing. Haven't tested Strix Halo myself. Did you consider DGX Spark as well? What is your target use case? Curious what feels suboptimal about llama.cpp + Vulkan so far.
4.
by Blue_Cosma 9mo ago
Our main driver and hypothesis was to work with regulated industry. We worked with a few large enterprise clients in defence and industry for R&D and IP use cases mostly. Our stack changes per project, adapting to client needs and infra
5.
Ask HN: Who's running local AI workstations in 2026?
9 points by Blue_Cosma 9mo ago|8 comments
6.
by Blue_Cosma 1y ago
Excellent point. How do you evaluate a developer’s ability to learn and integrate new tools (like AI assistants) into their workflow?
7.
by Blue_Cosma 1y ago
This is indeed during interviews (question updated) Thanks! I assume the 1-hour coding session is done live. From your experience, do candidates seem comfortable using AI tools as naturally as they would on their own? Do you also pay attent
8.
by Blue_Cosma 1y ago
AI coding tools (Cursor, Claude Code, etc.) are now part of most developers’ daily workflow, they speed up prototyping, planning, implementation but also change how we think and debug. I’m curious how you assess developers’ ability to lever
9.
Ask HN: How do you assess developers' AI-assisted coding skills in interviews?
5 points by Blue_Cosma 1y ago|10 comments
10.
Memory-Centric AI: SanDisk's High Bandwidth Flash(sandisk.com)
3 points by Blue_Cosma 1y ago|0 comments
11.
Open-Weight Agent from H Company(hcompany.ai)
2 points by Blue_Cosma 1y ago|0 comments
12.
Ask HN: Are embeddings too expensive for large datasets?
1 points by Blue_Cosma 2y ago|1 comments
13.
Show HN: Most Efficient Batch API for Open-Source and Custom Models(withexxa.com)
1 points by Blue_Cosma 2y ago|0 comments
14.
Show HN: Slash your LLM Inference Costs with Overnight Processing
5 points by Blue_Cosma 2y ago|1 comments
15.
Show HN: Slashing LLM Costs for Overnight Batch Inference
2 points by Blue_Cosma 2y ago|0 comments
16.
by Blue_Cosma 2y ago
Hi, Constant here. I'm the Chief Mobster! On hallucinations, larger models like Llama 3.1 70B still have a clear advantage, but they're also too costly to run at the moment on most API for those use cases. We built our API specifi
17.
by Blue_Cosma 2y ago
Post here if you have any comments or questions.
18.
Key Tech Job Market Trends and Hottest Skills: Insights from 84K HN comments(twitter.com)
4 points by Blue_Cosma 2y ago|3 comments
19.
by Blue_Cosma 2y ago
Let me know if you have any questions about the article
20.
Insights from 84,000 comments on "Ask HN: Who Is Hiring" using Llama3.1-70B-FP16(exxa.notion.site)
8 points by Blue_Cosma 2y ago|1 comments