Denis Remizov

CEO & Founder @SecuriLayer | Built enterprise API security in 40 days, solo | Also shipping TurboQuant (8.5x LLM compression) | ML engineer | 28

[Denis Remizov @remizov_dr]

@beginbot :)

[Securi Layer @SecuriLayer]

@SecuriLayer
is proud to join the
@Cloudflare
for Startups program with $10,000 in support & Enterprise Edge infrastructure! 🚀 Scaling real-time threat evidence for AI agents & enterprise workflows safely on Cloudflare. Thank you
@Cloudflare
team! 📷#Cybersecurity #AIAgents

[Securi Layer @SecuriLayer]

Big milestone for SecuriLayer 🔐

We've been accepted into the @TenableSecurity Alliance Program and
received a full NFR license for Tenable Vulnerability Management —
to build and validate our integration as an official tech partner
#cybersecurity #Tenable #startup #infosec

[Denis Remizov @remizov_dr]

@i_mika_el @SecuriLayer @zapier Привет, Использую бенчмарк безопасных промптов и логирование в теневом режиме для настройки порогов

[Securi Layer @SecuriLayer]

🚀 SecuriLayer is officially live on @zapier

Building AI agents? Add a security layer in 2 mins, zero code.

Drop us into your Zap to block Prompt Injections, Data Exfiltration & Malicious URLs before they hit your LLM.
▶️ https://t.co/YGWInaJz4O

[Securi Layer @SecuriLayer]

We just launched a native TAXII 2.1 server embedded directly into SecuriLayer. 🛡️

Now, Orbit chains & Arbitrum dApps can consume automated STIX2 threat feeds (Scam DNA) with cryptographic confidence scoring.

No more manual threat sharing. Standardized cyber defense for L2. 🛠️

[Securi Layer @SecuriLayer]

Backed by Optimism ASP Audit Grant
Want to collab? Cut the fluff. We have a powerful, live SDR Program (RevShare).
DM a concrete proposal:
Your value (Marketing/BD/Comm)
Your expectations
Early results = highest RevShare tier.

[Securi Layer @SecuriLayer]

Big milestone for SecuriLayer 🔥

Officially approved by OpenAI and added to their Plugin Store.

Now protecting #Web3‌‌ communities etc... Directly through ChatGPT.

Enterprise-grade threat intelligence, AI-powered, live on 12 platforms.

[Denis Remizov @remizov_dr]

🚀 TurboQuant-MoE v0.3.0 released!

• Up to 15.4× KV-cache compression
• Cross-layer delta + 3-bit PolarQuant

Serious VRAM killer for MoE models (Mixtral, DeepSeek, Qwen etc.)

https://t.co/b3H6uSVzWN

cc @tri_dao @vllm_project @TheBlokeAI

#LLM #MoE #KVCache #Quantization

[Denis Remizov @remizov_dr]

shipped TurboQuant while building SecuriLayer

8.5x LLM KV cache compression
0% quality loss
works today

256MB → 30MB
8.48x faster
same quality

free. MIT license.

https://t.co/Gp1IHvW62P

#LLM #AI #DevTools #OpenSource

[Denis Remizov @remizov_dr]

— this might be interesting. 📷 For now the account is fresh, so I’ll be posting updates, technical breakdowns and behind-the-scenes as I prepare for launch. Drop a 📷 if you want to follow the journey. #Web3Security #CryptoSecurity #AntiScam

[Denis Remizov @remizov_dr]

Hey X 👋 I’m a 28-year-old solo founder. In just 40 days I built SecuriLayer — a full-stack anti-scam protection system for Web3 communities. Covers Discord, Telegram, Slack + API + Browser Extension. No more drainers, pig butchering or homograph scams hitting your members. 📷

[Denis Remizov @remizov_dr]

This isn’t yet another simple bot or extension. It’s a real multi-surface shield built from scratch by one person. Project is in final pre-GA stage. Full launch is coming very soon. If you’re running a Discord/Telegram community or building in Web3 and tired of constant scams

[Denis Remizov @remizov_dr]

TurboQuant-MoE:8.5x KV-Cache Compression: 8.5x KV-cache compression for LLM inference https://t.co/9jC4rcsaj8 via @producthunt

[Denis Remizov @remizov_dr]

TurboQuant trending

biggest feedback: "why didn't this exist before?"

because:
1. hard math
2. people don't ship
3. I was angry about inference costs

3 things = one product

#ProductLaunch #Trending #LLM

[Denis Remizov @remizov_dr]

shipped TurboQuant while building SecuriLayer

8.5x LLM KV cache compression
0% quality loss
works today

256MB → 30MB
8.48x faster
same quality

free. MIT license.

https://t.co/Gp1IHvW62P

#LLM #AI #DevTools #OpenSource

[Securi Layer @SecuriLayer]

We built TurboQuant-MoE in 12 hours. The numbers surprised us too.

🧵 Thread:

1/ The problem: running Mixtral-8x7B requires ~90GB GPU RAM. Most teams can't afford that. We asked: what if compression didn't mean quality loss?

#TurboQuant #ML #github

[Denis Remizov @remizov_dr]

why 8.5x compression matters:

company: $10k/month inference, 8 GPUs, 500 users
with TurboQuant: $1.2k/month, 1 GPU, 4000 users

that's not optimization
that's game-changing

free. open source.

https://t.co/Gp1IHvW62P

#Infrastructure #Scaling