Cloud Security & Full Stack Developer
Chennai, India • LinkedIn • Portfolio • LeetCode

Security-minded software engineer specializing in automated cloud security, applied graph ML, high-concurrency backend pipelines, and AI system security. Driven by first-principles problem solving—moving beyond finding security gaps to building high-throughput, self-remediating infrastructure.

Engineered a multi-cloud IAM attack path analyzer using FastAPI, Neo4j, and Node2Vec anomaly detection.

• Ingests AWS & GCP IAM permission models into a Neo4j graph (13 node types) to execute Cypher queries across 29 privilege-escalation patterns.
• Combined deterministic rule checks with Node2Vec embeddings and Local Outlier Factor (LOF) graph-anomaly scoring; scanned a 1,000-principal environment in ~5s (~30s end-to-end) with zero false positives across test sets.
• Interactive Next.js graph rendering handles 391 nodes in ~1.5s.

Automated multi-cloud compliance and financial risk quantification engine built on hybrid RAG.

• Ingests live AWS/Azure telemetry into a hybrid BM25 + vector search RAG pipeline (LlamaIndex + ChromaDB) over 6 regulatory frameworks (232 controls across SOC 2, ISO 27001, NIST CSF 2.0).
• Automates control mapping and remediation drafting, cutting compliance triage from 30 → 6 minutes per finding (80% reduction).
• Integrates the FAIR framework (ALE) to quantify technical findings into executive exposure ($180K / INR 1.6 Cr).

Production-tested event-driven notification platform built for resilience and zero data loss under chaos conditions.

• Built using Django, Channels (ASGI), Celery, Redis, and PostgreSQL with SELECT FOR UPDATE SKIP LOCKED escalation sweepers and HMAC-signed acknowledgement links.
• Benchmarked via k6 on a single t3.micro: sustained ~30 events/s (560ms p95), 500 concurrent WebSockets, and ~1.0s p95 delivery.
• Maintained zero lost or duplicated messages across 1,300 chaos events using transactional outbox and jittered backoffs; fixed an ASGI sync-ORM leak, cutting HTTP 5xx rates from 6.88% to 0%.

Automated incident response engine utilizing unsupervised ML and Graph-based blast radius mapping.

• Trains an Isolation Forest pipeline over 30K real AWS CloudTrail log events across 7 behavioral features (13ms p95 inference latency).
• Maps detected anomalies inside Neo4j to evaluate 4-hop attack blast radius paths, gating automated Terraform remediation behind OPA/Rego policy checks to cut sub-35s MTTD/MTTR.

3-way financial reconciliation workbench built using Java 21 & Spring Batch.

• Processes 100K+ transaction rows at ~850 rows/s on an EC2 t3.micro using Spring Batch checkpointing and PostgreSQL SKIP LOCKED.
• Asynchronous ingestion pipeline (Next.js → Spring Boot → S3 → SQS) sustains ~120ms p95 API latency.

• Languages: Python, TypeScript, Java (21), SQL, Cypher, Bash
• Applied AI & ML: RAG Pipelines, LlamaIndex, ChromaDB, Hybrid Search (BM25 + Vector), OpenAI / LLM APIs, Node2Vec, scikit-learn (Isolation Forest, LOF)
• Backend & Systems: FastAPI, Django REST Framework, Django Channels (Daphne/ASGI), Spring Boot, Spring Batch, Celery, WebSockets, REST APIs
• Security & Cloud: AWS (EC2, S3, SQS, CloudTrail, GuardDuty), Azure, GCP, Docker, Terraform (IaC), OPA/Rego, OWASP ZAP, SonarQube, Trivy, Prowler
• Data Stores & Messaging: PostgreSQL, Redis (Pub/Sub & Caching), Neo4j, SQLite, AWS SQS

• Microsoft Certified: Azure Security Engineer Associate (AZ-500)
• Microsoft Certified: Security, Compliance & Identity Fundamentals (SC-900)
• Coursera / Professional: Security in Google Cloud Specialization | AWS Cloud Technical Essentials

• Open Source: Active contributor to security and cloud tooling repos (currently engaging with OWASP nettacker security modules & others).
• Current Focus: Building production-grade AI Agent guardrails, optimizing asynchronous RAG pipelines, and practicing LeetCode.
• Goal: Seeking SDE, Applied AI, or Cloud Security / DevSecOps engineering roles at product-first tech firms and GCCs.

Travelling • Music • Watching films • Tech Hackathons & CTFs