dmshirochenko Profile Banner
Shirochenko Dmitriy Profile
Shirochenko Dmitriy

@dmshirochenko

Followers
2K
Following
3K
Media
285
Statuses
3K

Software Engineer specializing in Python | Acing LeetCode for Optimal Solutions | Unbiased Coursera Course Reviews | 60+ Courses Mastered

Barcelona
Joined January 2011
Don't wanna be here? Send us removal request.
@dmshirochenko
Shirochenko Dmitriy
2 years
I completed Meta Back-End Developer! Check out my certificate #Coursera.
15
1
80
@dmshirochenko
Shirochenko Dmitriy
6 hours
The ‘scrape it all’ approach to building datasets is officially untenable; what comes next? This isn't just a privacy issue—it's a fundamental data integrity crisis for the entire field. We've focused intensely on model architecture, but this research underscores that the most.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
12 hours
What happens when a "financially unfeasible" VFX shot becomes 10x faster and cheaper to produce with generative AI? Netflix’s recent use of GenAI in a final production shot is a critical data point for the future of creative development. The bottleneck is no longer just budget or.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
18 hours
With so much focus on foundational models, are we overlooking the critical importance of the user interface and workflow integration in AI coding? The real battleground for developer adoption isn't just about having the most powerful LLM, but about creating the most seamless,.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
24 hours
What is the true technical debt of building on a foundation of indiscriminately scraped web data? The recent audit of the DataComp CommonPool dataset reveals a sobering truth: personally identifiable information isn't an edge case, it’s a core component of large-scale web.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
1 day
What if the most powerful infrastructure we build isn't physical, but a software layer over the hardware already in our pockets? The Android Earthquake Alert system is a profound example, converting millions of phones into a distributed, real-time seismic sensor network. This.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
1 day
How do we secure an ecosystem built on millions of low-cost, uncertified devices? Google's lawsuit against the 10-million-device BadBox 2.0 botnet is a direct confrontation with the security debt of open-source hardware. Malware pre-installed at the factory level shows the attack.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
2 days
What is the true cost of an insecure, open-source hardware ecosystem? The BadBox 2.0 botnet compromised 10 million smart devices, highlighting a massive supply chain vulnerability that goes far beyond simple ad fraud. This isn't just a failure of individual devices, but a.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
2 days
What if social media wasn't a product to be sold, but a protocol to be built upon? Jack Dorsey's $10M bet on the "and Other Stuff" collective is a crucial test for this very idea, directly challenging the platform-as-silo model. This push towards open protocols like Nostr is less.
0
0
1
@dmshirochenko
Shirochenko Dmitriy
2 days
How do you build for the long-term on a foundation you don't control? Broadcom's new invite-only VMware partner program is a massive disruption for the ecosystem, particularly for the small and mid-sized providers who built their businesses on it. This move highlights the.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
2 days
Is the "third workspace" a genuine productivity leap or the final erosion of our focus? Integrating Microsoft Teams with in-car cameras isn't just a new feature; it's a test case for ambient computing where the OS is the vehicle itself. The success of these systems won't be.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
3 days
What does it mean when a data-labeling giant lays off 14% of its staff just weeks after a massive strategic investment? This isn't a simple business correction; it's a powerful signal that the core of AI development is shifting under our feet. The focus is rapidly moving from.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
3 days
As the hype cycle cools, the data is stark: 80-90% of AI proofs-of-concept fail to scale. The real challenge isn't building a demo, but finding sustainable value in a landscape flooded with "AI novelty" projects. For developers, the most durable wins are coming from targeted.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
3 days
What happens when the tools we build to improve our workflow are used to eliminate it entirely? The report of King developers training the AI that replaced them is a stark case study in automation's double-edged sword. This isn't just about efficiency; it's a fundamental question.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
3 days
What happens when the training data itself becomes a persistent, costly bug? The issues with Google's Veo 3 generating gibberish subtitles are a perfect case study in the challenges of production AI. It's a stark reminder that impressive demos often hide a long tail of.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
4 days
What happens when a foundational model learns the wrong thing so deeply that you can't prompt your way out of it? Google Veo’s persistent, garbled subtitles are a classic example of training data artifacts becoming expensive, user-facing bugs. The problem isn’t the prompt, it’s.
0
0
2
@dmshirochenko
Shirochenko Dmitriy
4 days
Is the future of franchise-building a platform API? Roblox's new licensing portal for IP like *Stranger Things* signals a systemic shift, transforming user-generated content from a legal gray area into a core business strategy. This isn't just about fan games; it's the.
0
0
1
@dmshirochenko
Shirochenko Dmitriy
4 days
We spend our days optimizing code and model architectures, but what happens when the real bottleneck is the physical world? Meta reportedly using tents for data centers is a critical signal that the race for compute has become a brute-force construction and energy problem. The.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
4 days
Are AI coding assistants a productivity multiplier or a cognitive tax for experienced developers? A new study from METR provides a stunning data point, finding that AI tools actually made senior open-source devs 19% slower on real-world tasks. This suggests the bottleneck isn't.
1
0
0
@dmshirochenko
Shirochenko Dmitriy
5 days
How do we bridge the gap between AI's potential and its practical application in complex software projects? A new study reports experienced open-source developers were 19% slower when using AI tools, a stark contrast to their own perception of being 20% faster. This isn't an.
0
0
0
@dmshirochenko
Shirochenko Dmitriy
5 days
What happens when the builders get fed up with the infrastructure they’re forced to use? This is a powerful reminder that some of the most significant innovation doesn't come from a new AI model, but from directly solving tangible, frustrating problems. It's a classic engineering.
0
0
0