I'm a DevOps & Cloud Engineer (~3 years production infra) actively crossing the frontier into AI/ML engineering and ZK cryptography / Web3 infrastructure. I build systems with CI/CD, typed code, modular architecture, and real infrastructure — not prototypes. Currently operating across:

• 🔐 ZK Cryptography & Web3 — Groth16 provers, Circom circuits, ZK mixnets, on-chain governance, trusted setup ceremonies
• 🤖 AI/ML & LLM Systems — RAG pipelines, multi-agent orchestration, LLM-native DevOps, semantic retrieval
• 🗜️ Data Engineering — extreme-density vaults, ML-driven compression, encryption at rest, Merkle integrity
• ☁️ Cloud & DevOps — GCP/AWS-native, Kubernetes, Terraform, full CI/CD pipelines
• 🛒 Product Engineering — full-stack e-commerce platforms on GKE with Next.js + FastAPI

ZK Cryptography | Zero-Knowledge Sovereign Network

Jurisdictionally-agnostic, cryptographically-sovereign P2P mixnet. Groth16 verifier with Poseidon hashing, depth-20 Merkle circuit (1M+ member cap), pot28 Hermez trusted setup via GitHub Actions, Yggdrasil 200::/7 enforced at the socket layer, anonymous on-chain governance. Audit preparation documents complete. NixOS hardware testing in progress.

Rust Solidity Circom Groth16 Poseidon Yggdrasil NixOS

Data Engineering | Extreme-Density Encrypted LLM Data Vault

Compress-then-encrypt vault purpose-built for LLM training sets. MsgPack + Zstd-L22 + AES-256-GCM + SHA3-256 Merkle integrity. S3 and GCS backends. 39× compression ratio with full encryption and tamper detection at rest.

Python FastAPI MongoDB MsgPack Zstd AES-256-GCM S3 GCS

AI/RAG | Enterprise RAG with Content-Addressable Integrity

Enterprise RAG system with SHA-256 content-addressable snapshot hashing, two-stage retrieval pipeline (BM25 → Claude semantic reranker → SHA-256 integrity gate), phantom detector, and append-only audit log. Grounded retrieval with provable integrity.

Python LangChain BM25 Claude API FAISS SHA-256

Product Engineering | Premium Two-Wheeler Safety Accessories Platform

Full-stack e-commerce platform targeting Bengaluru — certification gate enforcing ECE 22.06 / Snell M2025 / SHARP 5-star standards only, daily freshness expiry rotation via Celery, GearCoin loyalty system, anti-counterfeit QR verification. Enterprise monorepo on GKE.

Next.js 14 FastAPI Celery GKE Terraform Kubernetes Prometheus Grafana

Platform | 15-Agent AI Software Factory

Ship production software without writing a single line. Deploys 15 Claude AI agents across Architecture → Dev → Testing → Security → DevOps. Every stage governed, every artifact immutable, every decision traceable.

Python FastAPI React PostgreSQL Redis Kafka Kubernetes Claude API

Automation | YouTube/Instagram Content Pipeline (novus.forge)

7-module Python automation system for faceless content: trend hunting → Claude-powered script generation → edge-tts voice synthesis → MoviePy video assembly → dual-platform upload. Powers the novus.forge content brand.

Python Claude API MoviePy edge-tts YouTube API Instagram API

Privacy | Biometric Privacy Shield

Biometric privacy app with liveness detection (Laplacian texture variance + EAR blink detection), hash-chained audit log, and full Vitest test suite.

TypeScript face-api.js Vitest Biometrics

AI Platform | CodeAct Multi-Agent Dev Workspace

FastAPI backend with four specialized agent classes (Planner, Coder, Tester, Debugger), Docker sandbox executor, WebSocket streaming, and Next.js Monaco-based IDE frontend.

Python FastAPI WebSocket Docker Next.js Monaco

Flagship | AI Agent for DevOps Incident Response

Local-first AI agent that diagnoses Kubernetes and CI/CD failures using real logs, cluster state, LLM reasoning, and RAG. Cost-aware AWS insights + automated Terraform remediation.

Python LangChain RAG FAISS Kubernetes AWS Apache 2.0

Research-Grade | Adaptive Universal Lossless Compression Engine

ML-driven universal lossless compression with multi-stage pipelines, intelligent strategy selection, explainable RAG reasoning, and live evaluation benchmarks.

Python ML RAG Compression Pipelines MIT

Production-Minded | Nano-Scale LLM System

Clean-room open-source nano LLM — training, RAG augmentation, tool use, evaluation harness, live demos. Grounded generation through retrieval and deterministic pipelines.

Python HuggingFace Transformers LoRA RAG Apache 2.0

IoT Security | Industrial Edge Security Platform

Real-time, safety-first industrial IoT security system. Raspberry Pi-ready, fully Dockerized, production-grade open-source.

Python Raspberry Pi Docker MQTT Apache 2.0

Research-Grade | Multi-Document Intelligence

Full-stack RAG QA across multiple documents. PDF/DOCX/TXT ingestion, vector DB retrieval, LLM-powered answers, explainable reasoning, evaluation metrics.

Python LangChain FAISS ChromaDB FastAPI Apache 2.0

AWS Serverless | Lambda + Step Functions Image Pipeline

Serverless image recognition using AWS Lambda, Step Functions orchestration, and Amazon Rekognition.

JavaScript AWS Lambda Step Functions Rekognition Apache 2.0

AWS DevOps | Full CI/CD Deployment Pipeline

End-to-end AWS deployment pipeline: CodePipeline → S3 → CodeDeploy on Linux. Zero-downtime deployments.

HTML AWS CodePipeline S3 CodeDeploy Apache 2.0