Building systems that scale  ·  Automating what can be automated  ·  Learning by shipping

{
  "name"      : "Saravana Priyan",
  "role"      : "Backend AI Automation Engineer",
  "degree"    : "B.Tech, Artificial Intelligence & Data Science",
  "interests" : ["Backend Engineering", "Workflow Automation", "Cloud Infrastructure", "AI-Integrated Systems"],
  "mindset"   : "Pick ambitious projects → break them down → learn along the way",
  "status"    : "Building in public 🚀"
}

I design backend systems that automate operational work - deployments, service orchestration, infrastructure workflows - and integrate AI as a working component inside that automation, not as a standalone feature. Systems that hold under pressure, APIs that don't break, pipelines that run reliably.

Visual workflows for build, test, deploy, and infrastructure operations - built backend-first.

MicrOps gives engineering teams a visual interface to design and execute DevOps pipelines - without writing boilerplate config from scratch. Behind the clean UI sits a backend engineered for reliability: workflow orchestration, cloud integrations, and automation pipelines tied together with modern development practices.

 ┌──────────────┐     ┌──────────────┐     ┌──────────────┐     ┌──────────────┐
 │    Build     │────▶│     Test     │────▶│    Deploy    │────▶│  Infra Ops   │
 └──────────────┘     └──────────────┘     └──────────────┘     └──────────────┘
        │                    │                    │                    │
        └────────────────────┴──────── Orchestrated by MicrOps ──────┘

Core Concepts: Workflow Orchestration · Cloud Integrations · Automation Pipelines · Backend-First Architecture

Stack: Node.js · Jenkins · AWS EC2/S3/ECR · Docker

→ View Project
→ View Repo

Speak naturally. Get placement guidance back as streamed speech.

Nephele connects microphone audio, speech recognition, a conversational LLM, and text-to-speech over a WebSocket voice pipeline. Built for interview prep, resume tips, and placement questions, with live transcripts, interruptible responses, and a React interface that starts listening only when you choose to talk.

Core Concepts: Streaming Audio · WebSockets · Speech-to-Text → LLM → Text-to-Speech · Interruptible Responses

Stack: Python · FastAPI · Deepgram Nova-2 / Aura · OpenAI GPT-4o-mini · React · TypeScript · Vite

Built by The Cloud PeP Students; tech leads: Saravana & Joshua.

→ View Project
→ View Repo

Selected contributions, with merged work separated from proposals still under review. Status checked on 15 October 2026.

Every project I've shipped has taught me something a tutorial never could - about trade-offs, system behaviour under pressure, and what it means to actually engineer software. I don't collect courses. I collect shipped things.

$ cat learning_progress.log

  Advanced Backend Engineering   ████████░░░   actively building
  System Design                  ██████░░░░░   deepening
  Kubernetes                     ████░░░░░░░   exploring
  Platform Engineering           ████░░░░░░░   exploring
  Distributed Systems            ███░░░░░░░░   foundations
  Data Structures & Algorithms   █████░░░░░░   practising daily

_{Built from scratch, not a template.  ·  Just like my projects.}