AI-Powered Security API Gateway

A stateless, high-throughput security gateway that detects and mitigates malicious traffic in real time using an event-driven architecture, enforcing gateway-level protection before requests reach backend services.

Key Design Decisions:

• Problem → Detect and block malicious clients without increasing request-path latency or consuming backend resources
• Solution → Fire-and-forget Kafka event pipeline with Redis-backed blacklist enforcement, Strategy Pattern-based threat detection, and AI-assisted analysis

Highlights:

• 🪪 Cryptographically signed Visitor ID tracking for secure client identification
• ⚡ Kafka-based asynchronous threat analysis with fully decoupled security services
• 🎯 Strategy Pattern-driven detection engine for rate abuse, bot activity, and behavioral anomalies
• 🔄 Circuit Breaker optimization to terminate analysis immediately after threat detection
• 🚫 Dynamic Redis-backed blacklisting with configurable time-based enforcement
• 🤖 Ollama LLM integration for AI-assisted threat classification when rule-based strategies fail
• 📊 Centralized security logging and historical context analysis

Consent-Driven Distributed Identity Platform Alpha 🚧

A microservices-based KYC orchestration system that eliminates redundant verification across institutions while enforcing secure, consent-based data access.

Key Design Decisions:

• Single Source of Truth → One verified identity reusable across organizations
• Consent → Explicit user authorization required before any data access
• Event-Driven → Asynchronous communication using Kafka
• Security → JWT authentication, RBAC, and encrypted data storage
• Lifecycle → Automated renewal, revocation, and update workflows

Highlights:

• 🔐 Consent-based identity sharing with strict access control
• ⚡ Event-driven microservices using Apache Kafka
• 🏛️ Multi-institution KYC orchestration
• 📊 Strong data governance and trust boundary enforcement

AI-Powered Repository Understanding & Architecture Visualization System

Intelligent codebase analysis platform that accepts Git repository URLs, performs AST-driven source code analysis, constructs dependency graphs, and enables natural-language interaction with complete software architectures.

Key Design Decisions:

• Architecture Mapping → AST-based parsing to identify file dependencies, imports, and structural relationships
• Knowledge Retrieval → RAG pipeline for repository-wide semantic search and architecture-aware question answering
• Visualization → Interactive dependency graph generation for large-scale codebase exploration

Highlights:

• 🔍 Automated Git repository cloning and codebase indexing
• 🌐 Interactive file-to-file dependency graph visualization
• 🧠 RAG-powered architecture intelligence and repository Q&A
• 📂 File location tracing, dependency discovery, and workflow explanation
• ⚡ FastAPI backend integrated with Anthropic AI for contextual code understanding

Demo:
🔗 https://drive.google.com/drive/folders/1FaOFA_c6KlkVEYwn5Hyu9Hnfw-ba03CM?usp=sharing

Automated Threat Detection Pipeline

Static analysis + ML pipeline to classify polymorphic malicious payloads hidden within PDF structures.

Key Design Decisions:

• Innovation → Custom structural parsers to identify high-entropy object streams in malformed PDFs
• Accuracy → XGBoost ensemble model for classifying zero-day polymorphic variants

Highlights:

• 🧠 XGBoost + TensorFlow classification pipeline
• 🔬 Custom PDF structure parser for malformed documents
• 📈 High-entropy stream detection