Sandijs Aploks's Avatar

Sandijs Aploks

@sandijsaploks.bsky.social

Neural Networks programmer. Backends programmer. Could help with MLOps and parallelism programming in neural networks training and inference. Learned testing backends for serious.

28 Followers  |  9 Following  |  54 Posts  |  Joined: 08.12.2024  |  1.381

Latest posts by sandijsaploks.bsky.social on Bluesky

Preview
AI Agents Programming Here are some important concepts to think about when developing AI Agents.

Important aspects when programming AI Agents:
www.linkedin.com/pulse/ai-age...

05.08.2025 08:28 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

How would Enterprise AI Agent look like? #ArtificialIntelligence

04.08.2025 10:46 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

I was thinking about BERT Transformers vocabulary, subwords, tokens, and Chinese language today.

How efficient are LLMs with Chinese language?

26.07.2025 21:48 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

LLMs need Thinking Language - every knowledge piece gets compiled to it.
Every question gets compiled to it.
We need to break free from Humans languages shackles. #neuralnetworks #artificialintelligence

26.07.2025 14:20 β€” πŸ‘ 2    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0
Preview
GitHub - google-gemini/gemini-cli: An open-source AI agent that brings the power of Gemini directly into your terminal. An open-source AI agent that brings the power of Gemini directly into your terminal. - google-gemini/gemini-cli

β™₯️ Google will get a lot of proprietary code examples to train their AI via Gemini CLI for terminals?
Will the companies have control over what programmers will upload to the Google servers ??? github.com/google-gemin...

26.06.2025 23:36 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

I’m seeking custom LLM / SLM / Transformers Developer and DevOps job.

I am low level Neural Networks Developer myself.

Currently I am creating Synthetic Dataset and Training and Fine-Tuning BERT Transformer.

I'm experienced Backend Developer & DevOps for Cloud and Kubernetes.

26.06.2025 22:50 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
Zero-Downtime Database Migration and Rollback In most cases, Backend downtime during the data and software Upgrade or Rollback is not acceptable. In some cases, data volume is too large for various data migration methods, and adding Canary releas...

Zero-Downtime Database Migration and Rollback
www.linkedin.com/pulse/zero-d...

12.06.2025 11:10 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Cloud Infrastructure Changes Transaction with automatic Rollback How to create cloud infrastructure changes transaction which commits or aborts just like in some database transaction? This transaction happens in longer time period, and includes many components incr...

Cloud Infrastructure Changes Transaction with automatic Rollback
www.linkedin.com/pulse/cloud-...

28.05.2025 14:53 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
How to automate AWS EKS Service Hot Fixes deployment or rollback? AWS Serverless Lambda Functions allow sophisticated DevOps scripts programming. It is possible to achieve more than with Terraform, Ansible, or even Ansible Server.

How to automate AWS EKS Service Hot Fixes deployment or rollback (DevOps)?
www.linkedin.com/pulse/how-au...

19.05.2025 11:19 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

Did you know that LLM companies know which people are the most useful ones to build next generation USA?
The questions and the topics they work on allow to spot exceptional individuals.
Both USA and China already have info about strategic people.

17.05.2025 15:58 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

I cannot get grants like you do. I pay for the science from my own pocket. Just like in those times when I was working on the Earth Space Agency project.
You can hire me too to work with MLOps / AIOps.

10.05.2025 10:14 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
How to make High Precision Neural Networks This is the summary of my mini-articles in my journey to learn High Precision Neural Networks programming. What is High Precision Neural Networks? Imagine Time Series and predicting numbers like 13.

I need your help to continue writing Neural Networks articles and to develop easy to learn source code examples: lnkd.in/d4rh83q7
To support me you can send money to this EU bank account: LV26HABA0551001470845 SANDIJS APLOKS
Current activity: Neural Turing Machines.

10.05.2025 10:11 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Post image

DeepSeek can help to better understand Neural Networks scientific papers.
It even did understand from which paper was this screenshot from.

10.05.2025 07:59 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Trump is taking away power from certain social type people who got that with Life Hacking which is a crime.

05.05.2025 20:55 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Julia is incredible at least for Neural Networks and Smart City Computer programmers.

05.05.2025 16:25 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

I am looking for Backend Developer and DevOps job.
Search for my profile on Linkedin.

30.04.2025 07:02 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Could Vibe Coders replace Amazon or Facebook?
That false euphoric hope is clearly visible in Latvia too.
The answer is NO.
One has to learn several lives in the row to become like Jeff, Elon, or Mark.
It means learning several professions, not just sloppy programming.
Practice.

25.04.2025 09:21 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

How to visualize Cloud Infrastructure changes time step after time step?

15.04.2025 08:35 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
Asynchronous and Durable 2 Phase Commit Here is proposal how classical XA Transactions 2 Phase Commit could be made asynchronous and durable to survive the crashes and network outages without causing deadlocks or loosing transactions. Schem...

Asynchronous and Durable 2 Phase Commit:
www.linkedin.com/pulse/asynch...

09.04.2025 11:51 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

How to create a Specialized Brain by using Low Level Neural Networks?
It is like in the nature - brain blobs and their communication.
Brain blobs architecture which outputs computed result.

24.03.2025 13:49 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Propaganda Warriors are attacking Musk and Trump.
Good question is how this all will end, because such attacks are illegal.
Once the government stops supporting it, criminal penalties start to apply.
Western usage of Propaganda against its own people must be stopped.

23.03.2025 09:41 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

When using Embedded Neural Networks to produce input for other Neural Network training Loss function extra punishments, it is possible to use Monte Carlo Dropout + Activation Neural Network to let through only Clear Signal which in Loss function would work mostly correctly.

21.03.2025 10:06 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Neural Networks programmers should try Qwen Max and its Reasoning.
What Chinese colleagues are using to train it?
Chinese robot makers private source code ???!!!

17.03.2025 21:40 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Reading DeepSeek, Grok, or Qwen REASONING process which happens before giving you a solution for some Neural Networks problem is more valuable than the answer itself.
I was shocked to learn perspectives and approaches I didn't know about.
My next mini-publication will be simply mind blowing.

13.03.2025 20:21 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

How to predict stock prices better by using 1K factors?

22.02.2025 17:18 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

I am looking for new backends developer job.
I can help with low level neural networks programming too.
Check my Linkedin profile for details.

11.01.2025 19:14 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Will AI kill huge programmers outsourcing companies like Accenture?
Accenture ~ 738,000 employees
Tata Consultancy Services (TCS) ~ 614,000 employees
Deloitte ~ 416,000 employees
EY (Ernst & Young) ~ 365,000 employees
Capgemini ~ 340,000 employees
Cognizant ~ 341,300 employees
Infosys ~ 335,000 empl

10.01.2025 09:44 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Governments should think about AI quotas per year.

Otherwise it will be about millions of unemployed IT specialists out there.

Transition programs must be created as all those super intelligent people would be a major gain for every business.

07.01.2025 20:17 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

I am writing High Precision Neural Networks framework to be used in Smart Cities type AI.

01.01.2025 22:11 β€” πŸ‘ 5    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
arXiv.org e-Print archiveopen searchopen navigation menucontact arXivsubscribe to arXiv mailings

Remember, high accuracy of the Neural Network (98% +) is important.

That means when you are writing an arxiv.org article, do not hide accuracy metrics, do not replace it with other metrics which make the article look much better.

31.12.2024 15:04 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

@sandijsaploks is following 9 prominent accounts