◇ identity
I offer all my experience wherever it is needed. Whether you need advice on securing a corporate network or employee communications in a high-risk area, conducting a risk analysis or preparing crisis scenarios, or “just” develop an application or dream software product, you’ve come to the right place.
“one way to show respect for people in everyday life is a well done job”
Be safe even from NSA
I have been working in cyber-security & cyber-defence since 2010, when I entered this field with my first clients and expanded my scope to include services in this area. Over time the portfolio has grown and I am now able to offer a wide range of cyber security / defence services — for individuals, companies and people who can’t afford a mistake.
use ip_spoofing::{self, RawSocket, ReusablePacketWriter};
fn main() -> ip_spoofing::Result<()> {
let socket = RawSocket::new()?;
let mut writer = ReusablePacketWriter::new();
socket.send_fake_udp_packet(
&mut writer,
[8, 8, 8, 8],
1234,
[127, 0, 0, 1],
5678,
b"hey",
64,
)?;
Ok(())
}Make your ideas real
I’ve been developing software for more than 14 years, and in that time I’ve created and delivered many diverse applications for desktop, web and mobile — and been partly involved in many others.
I’ve been part of interesting government contracts, collaborated on security-architecture design for companies like Proton AG (Protonmail), and worked in a war zone — extreme conditions and pressure, where a mistake doesn’t just crash the app, in the worst case it costs human life.
private unsafe double CalculateSum(int count)
{
double sum = 0;
double* values = stackalloc double[count];
for (int i = 0; i < count; i++)
{
sum += values[i];
}
return sum;
}Make machines work for your business
These days I help teams put LLMs and agents to actual work — wired into the software you already run, not as a separate toy. From small in-product copilots to background agents that move data, draft documents or triage tickets, the focus is on what the system has to deliver, not what model is fashionable this week.
I also run research engagements: where in your team’s day does an LLM actually shorten the loop, and where does it just add risk? With a background in cyber-security I treat prompt injection, data exfiltration and model trust as first-class concerns — so the integration ships in a way you can still defend.
from anthropic import Anthropic
client = Anthropic()
def triage(ticket: str) -> str:
msg = client.messages.create(
model="claude-sonnet-4-5",
max_tokens=512,
system="Classify support tickets.",
messages=[{"role": "user", "content": ticket}],
)
return msg.content[0].text◇ trusted by
◇ regions deployed
◇ get in touch
Send a brief. I read every message myself. Encrypted communication available on request.
◇ end of file
© 2026 pitri.net · all bytes reserved