Parth Garg

Smart grid monitoring system using basic ML classification (on SCADA data) to predict LT line faults in real-time with 85% accuracy • Dual-interface system with admin dashboard, Gemini-powered AI chatbot, and multi-channel alerts (Twilio SMS, Nodemailer); reduced fault response time by 60% • Applied data analysis techniques on SCADA time-series data for feature engineering and fault pattern identification • Built RESTful APIs with Express.js and Flask; designed modular backend for seamless integration of ML inference and alert dispatch services

Zero-DevOps removes the DevOps layer entirely. You connect a GitHub App, pick a repository, and the platform handles the rest — from immutable, reproducible builds to artifact storage. The system is built on four deliberate architectural bets: - Builds are immutable, not moving targets. Every build is keyed to an exact commit SHA resolved at trigger time — the worker never builds a shifting branch tip, so every run is reproducible and queryable. - The queue is the boundary. A Go API server only validates and persists; a separate Go worker consumes jobs via RabbitMQ. Neither side knows about the other's internals, so each can scale, fail, and evolve independently. - Contracts are versioned files, not conventions. The deploy.jobs V1 message is a JSON Schema both services compile against, with conformance tests enforcing the contract. - Detection replaces configuration. The worker auto-detects the framework and package manager and renders a Dockerfile from templates — you only ship your own Dockerfile if you want to. The result: a developer connects a repo, triggers a build, and gets a container image artifact in Cloudflare R2 — with zero DevOps work in between. This is still under development