About

AI SECURITY · NETWORK SECURITY ENGINEER · APPLIED AI

Welcome to pawanbk.io

I'm Pawan Bishwokarma, a cybersecurity professional exploring the intersection of AI, security, network defense, and applied machine learning.

This site is where I share what I’m learning, building, and thinking about as AI continues to reshape how we work, defend systems, and understand technology.

Why this site exists

I created this site for anyone who wants to learn about AI concepts, follow along with my AI projects, and hear my personal perspective on new developments in the field.

Some posts will be technical and hands-on. Others will be more reflective : what I’m noticing, what I’m questioning, and what I think matters as AI becomes more connected to cybersecurity, infrastructure, and everyday tools.

What I write about

  • AI concepts — clear explanations of ideas like agents, RAG, embeddings, model context, and AI workflows
  • AI security — risks, attack paths, defensive patterns, and how security changes in the AI era
  • Network defense — detection logic, intrusion detection, traffic analysis, and security monitoring
  • Applied AI projects — systems I build, test, break, improve, and document
  • Personal perspective — thoughts on new tools, trends, and where I think the field is heading

Projects I’m building

One of my current projects is SentinelMesh, an AI-powered IoT intrusion detection system that combines a CNN classifier, LSTM anomaly detection, and an edge-based security architecture.

I use projects like this to connect theory with practice: not just asking what is this concept? but how would this actually work in a real system?

Start here

A good place to begin is the SentinelMesh series, where I walk through how I designed and built an AI-powered intrusion detection system for IoT environments.

Feel free to connect with me on LinkedIn. I am always open to connecting and discussing AI, security, and network defense.

Welcome to pawanbk.io!

Focus areas

Click a topic to read more.

Prompt injection, data leakage, and guardrail bypass research — how models are abused in production and what actually stops it.

Autonomous agents multiply attack surface: tool misuse, privilege escalation, and unsafe chaining across APIs and memory.

Segmentation, detection, and resilient architecture — containing breaches before they become enterprise-wide incidents.

STRIDE-style analysis adapted for ML pipelines, RAG systems, and hybrid cloud/on-prem deployments.

Telemetry and policy layers that make AI workloads auditable — the foundation behind SentinelMesh.

Turning academic ideas into reproducible experiments and write-ups other engineers can use.