Rahul Sumbly
Security • Anti-Fraud • AI Agents • Offensive & Defensive AI. Specializing in scalable intelligence that protects the digital frontier.
I am Rahul Sumbly, a Senior AI Engineer dedicated to the intersection of machine learning and cybersecurity. In an era where AI can be both the shield and the sword, I focus on building resilient systems that anticipate malicious intent and automate the defense of critical digital infrastructure.
ML + LLM-driven log analysis and behavior baselining on a serverless GCP architecture, processing 100M+ logs per day. Autoscales seamlessly from a 1,500 RPS baseline to 3,000 RPS at peak load, while expanding log source coverage by 300%.
100M+ Logs / Day · ML · LLM · Security Operations · GCP · Serverless
24/7 autonomous agent using custom security playbooks. Saved 1000+ analyst hours/month and reduced MTTA by 80%.
1000h+ Saved / Month · Agents · Automation · Python
Enterprise-grade no-code platform enabling non-technical users to build and deploy sophisticated AI agents in seconds. Simplifies agentic workflows with a single-click orchestration engine.
1-Click Agent Deployment · No-Code · Generative AI · Platform
End-to-end automated adversarial testing framework for AI agents. Simulates complex multi-stage attacks to stress-test LLM guardrails and agentic logic.
100% Agent Pentesting Automated · Red Teaming · Cybersecurity · AI Safety
Distributed multi-agent AI system performing deep, collaborative vulnerability assessments with highly optimized discovery engines.
90% faster recon and discovery · Multi-Agent · Pentesting · Scaling
High-precision systems for No-Trade, Short-Trade, and Payment Fraud detection using behavioral analytics and ML.
65% more cases captured · Finance · Anti-Fraud · Analytics
· 8 min read
I recently interviewed candidates for an AI Engineer role. Here's what actually helped people, what quietly hurt them, and a few things I wish someone had told every candidate before they joined the call.
· 26 min read
How Anthropic, OpenAI and Google price tokens, why cached and batched tokens are cheaper, how to route for cache hits, and the easy-to-miss bugs (timestamps, unsorted JSON, reordered tools) that quietly wreck your hit rate.
Birla Institute of Technology and Science (BITS) Pilani, B.Tech in Computer Science (2021 – 2025)