Nazar Vozniak
@vozniaknazar.bsky.social
Full-stack developer and builder who works at the intersection of software, AI, and real-world business. He designs systems end-to-end — from backend architecture and data pipelines to SEO, analytics, and product strategy.
We built ArXiv Intelligence to help you: 🏃♂️ Process dense papers in seconds. 🤝 Collaborate with your team in one workspace.
Ideal for researchers and entrepreneurs who need to move faster than the arXiv AI papers.
arxivai.iip.com.ua
Ideal for researchers and entrepreneurs who need to move faster than the arXiv AI papers.
arxivai.iip.com.ua
February 2, 2026 at 9:01 AM
We built ArXiv Intelligence to help you: 🏃♂️ Process dense papers in seconds. 🤝 Collaborate with your team in one workspace.
Ideal for researchers and entrepreneurs who need to move faster than the arXiv AI papers.
arxivai.iip.com.ua
Ideal for researchers and entrepreneurs who need to move faster than the arXiv AI papers.
arxivai.iip.com.ua
Moltbot (www.molt.bot) is in the “local + action-taking” camp. Sounds great (privacy, less context switching, real execution), but raises the real questions: setup friction, permissions, audit trails, and how much power you’ll actually hand an agent.
#AI #AIAgents #Automation #Privacy #LocalFirst
#AI #AIAgents #Automation #Privacy #LocalFirst
Moltbot — Personal AI Assistant
Moltbot — The AI that actually does things. Your personal assistant on any platform.
www.molt.bot
January 28, 2026 at 3:10 PM
Moltbot (www.molt.bot) is in the “local + action-taking” camp. Sounds great (privacy, less context switching, real execution), but raises the real questions: setup friction, permissions, audit trails, and how much power you’ll actually hand an agent.
#AI #AIAgents #Automation #Privacy #LocalFirst
#AI #AIAgents #Automation #Privacy #LocalFirst
A lot of people ask if synthetic data is 'real enough' to train a defense system. The answer lies in the Provenance Hypergraph.
www.linkedin.com/pulse/bridgi...
#AI #MachineLearning #InformationWarfare #GNN #NLP #CyberDefense #Ukraine #SyntheticData #OSINT
www.linkedin.com/pulse/bridgi...
#AI #MachineLearning #InformationWarfare #GNN #NLP #CyberDefense #Ukraine #SyntheticData #OSINT
Bridging the Data Gap: A Technical Deep-Dive into Synthetic Data for Information Defense
Last week, we shared a post saying that we are working on a new multi-modal AI framework designed to detect Russian disinformation. Today, we want to address the technical engine driving this system: ...
www.linkedin.com
December 28, 2025 at 1:57 PM
A lot of people ask if synthetic data is 'real enough' to train a defense system. The answer lies in the Provenance Hypergraph.
www.linkedin.com/pulse/bridgi...
#AI #MachineLearning #InformationWarfare #GNN #NLP #CyberDefense #Ukraine #SyntheticData #OSINT
www.linkedin.com/pulse/bridgi...
#AI #MachineLearning #InformationWarfare #GNN #NLP #CyberDefense #Ukraine #SyntheticData #OSINT
Lately I’ve been doing something both energizing and exhausting: monitoring the newest AI papers on arXiv every single day.
There’s something addictive about watching ideas emerge in real time — small conceptual shifts, strange experiments, unexpected architectures.
There’s something addictive about watching ideas emerge in real time — small conceptual shifts, strange experiments, unexpected architectures.
December 10, 2025 at 10:42 AM
Lately I’ve been doing something both energizing and exhausting: monitoring the newest AI papers on arXiv every single day.
There’s something addictive about watching ideas emerge in real time — small conceptual shifts, strange experiments, unexpected architectures.
There’s something addictive about watching ideas emerge in real time — small conceptual shifts, strange experiments, unexpected architectures.
Currently reading: DeepSeekMath-V2 Towards Self-Verifiable Math Reasoning
The potential in this specific architecture seems quite big. We are finally seeing a shift from LLMs that just "predict" the next word to systems that verify their own logic before outputting result.
The potential in this specific architecture seems quite big. We are finally seeing a shift from LLMs that just "predict" the next word to systems that verify their own logic before outputting result.
December 6, 2025 at 2:48 PM
Currently reading: DeepSeekMath-V2 Towards Self-Verifiable Math Reasoning
The potential in this specific architecture seems quite big. We are finally seeing a shift from LLMs that just "predict" the next word to systems that verify their own logic before outputting result.
The potential in this specific architecture seems quite big. We are finally seeing a shift from LLMs that just "predict" the next word to systems that verify their own logic before outputting result.
A marketing manager with zero coding skills used ChatGPT to expose critical vulnerabilities in a Fortune 500 website in just 15 minutes 🚨
AI found exposed APIs, weak authentication & SQL injection points. The technical barrier to cybercrime just disappeared 💀
We're not ready for what's coming 🔒
AI found exposed APIs, weak authentication & SQL injection points. The technical barrier to cybercrime just disappeared 💀
We're not ready for what's coming 🔒
November 30, 2025 at 11:03 AM
A marketing manager with zero coding skills used ChatGPT to expose critical vulnerabilities in a Fortune 500 website in just 15 minutes 🚨
AI found exposed APIs, weak authentication & SQL injection points. The technical barrier to cybercrime just disappeared 💀
We're not ready for what's coming 🔒
AI found exposed APIs, weak authentication & SQL injection points. The technical barrier to cybercrime just disappeared 💀
We're not ready for what's coming 🔒