Subhendu S.

About Subhendu S.

Driving a unified AI-native strategy across cloud security and intelligent automation, 4 initiatives — hybrid knowledge base search, IaC blueprint embedding in Windsurf, MCP interface for AWS Security Hub, and CI/CD compliance validation via AWS Bedrock — are live in production, delivering measurable reductions in misconfigurations and compliance findings. Alongside these, 3 cutting-edge research initiatives in intent preservation, intelligent context compaction, and LLM vulnerability guardrails are actively shaping the next frontier of secure, grounded, and cost-efficient agentic AI workflows at scale.1. Intent-Based Knowledge Base with Hybrid SearchAI-driven hybrid engine fusing BM25 lexical precision and semantic embeddings, delivering intent-aware ranked results with LLM-powered fidelity scores for high-confidence, contextually grounded responses.2. Embedded Blueprint Knowledge Base (Windsurf)Generative AI embedded into Windsurf IDE, empowering DevSecOps teams with AI-assisted, policy-aware IaC code generation to autonomously reduce misconfigurations and eliminate compliance findings at the source.3. MCP Interface for AWS Security HubAI-powered MCP server transforming AWS Security Hub findings into queryable AI context, enabling autonomous agents to intelligently surface, triage, and remediate cloud security posture in natural language.4. CI/CD Compliance Validation via AWS Bedrock Prompt FlowsAWS Bedrock prompt flows embedded into CI/CD pipelines to autonomously validate IaC code against CIS, FSBP, and PCI-DSS — shifting compliance left before anything reaches production.5. Intent Preservation in Multi-Tier Agentic AIGenerative AI-native intent propagation layer encoding goals and constraints as structured signals, keeping every agent semantically anchored to the original request regardless of reasoning chain depth.6. Intelligent Context Compaction for Agentic WorkflowsLLM-powered summarization to dynamically compress long-running agentic context, slashing token overhead, cutting inference costs, and sustaining coherent reasoning without losing critical task state.7. LLM Vulnerability Guardrails for Agentic AIAI-native guardrails continuously auditing agentic prompts, tool calls, and inter-agent communications against organizational policies — neutralizing prompt injection, data leakage, and misalignment risks before they propagate.

Experience

1. Team Lead

   Swift

   May 2005 — Present · US

   Built GenAI-powered cloud knowledge base using Bedrock and OpenSearch VectorDB, improving troubleshooting, onboarding, and internal documentation access.• Applied GenAI with MCP-compliant server to automate infrastructure compliance checks—enhancing audit readiness, governance, and alignment with PCIDSS, CIS, and AWS best practices.• Led AWS Cloud Landing Zone setup for 40+ apps, creating secure, scalable, multi-account environments with automation (CloudTrail, Config, Macie, Inspector), network segmentation, observability, and compliance guardrails embedded into CI/CD.• Migrated 20+ legacy enterprise apps to AWS using ECS Fargate, Aurora Serverless, Lambda, and API Gateway—boosting performance, resiliency, and reducing operational costs.• Built DevSecOps pipeline integrating security scans and compliance controls, reducing production vulnerabilities by 79% and accelerating deployments.• Implemented Zero Touch Deployments with CloudFormation, CDK (Python), and Terraform—achieving 99.99% uptime and enabling 24/7, PCIDSS-compliant releases.• Enforced Zero Trust security using IAM, AD Federation, SCPs, encryption, Config Rules, Security Hub, GuardDuty, Macie, XSOAR, and SIEM—improving overall security posture by 79%.• Established observability using CloudWatch, X-Ray, and Synthetic Canary; centralized logs in Elasticsearch; integrated alarms with ServiceNow to improve MTTR and visibility.• Partnered with AWS, Capgemini, and TCS on SoWs for mobile CI/CD, SaaS AI assessments, and hybrid monitoring—ensuring on-time and within-budget delivery.• Led cloud-wide penetration testing with NCC Group; achieved 78% SIP score in AWS-hosted security assessment.• Migrated monolithic app to microservices—cut infra cost by 50%.• Built REST/API-based event-streaming platform for 200+ apps—saved $1M in legacy license costs.• Migrated WebMethods to RedHat Fuse ESB.• Modernized Ops Management using REST APIs, HTML5 frontend, and microservices backend.

Education

• Indian Institute of Technology (Banaras Hindu University), Varanasi

  B Tech, computer

  1993 — 1997