Hello, I'm

Sagar Kharbikar

Backend Developer

I build the systems behind the screen — designing APIs, automating workflows, and engineering AI-powered backends that work in production.

View My Work Get In Touch
Node.js Primary Runtime
Python AI & Automation
MongoDB Data Layer
saar.js 
const express = require('express');
const router = express.Router();

// Document lifecycle: SUBMITTED → APPROVED
router.post(
  '/documents/:id/approve',
  authenticate,
  authorize('verifier'),
  async (req, res) => {
    const doc = await
      Document.findById(
        req.params.id);
    await doc.transition(
      'APPROVED');
    res.json({ status: doc.state });
  });
// about

How I Think About Backend

I approach backend development as an engineering discipline — not just writing code that works, but building systems that handle failure gracefully, scale without rewriting, and remain maintainable six months later.

My work spans Python and Node.js ecosystems — FastAPI for high-performance APIs, Express.js for MERN-stack backends, MongoDB for flexible data modeling, and OpenCV + Twilio for AI-driven automation. I've built everything from voice-controlled AI assistants to award-winning ERP portals.

I think in terms of data flow: how a request enters the system, gets validated, hits the service layer, interacts with the database, and returns a well-structured response. Every layer has a responsibility, and I keep them separated.

⚙

API-First Design

Define contracts before writing implementation. Clear endpoints, proper HTTP methods, consistent response structures.

🗃

Data Integrity

Enforce constraints at the database level. Use transactions where consistency matters. Index strategically.

🔒

Security by Default

Authentication and authorization aren't afterthoughts. Role-based access, input validation, and secure session handling from day one.

// projects

What I've Built

Real systems — API design, AI pipelines, automation, and backend architecture built independently from scratch.

🏆 1st Prize — College Level 01

Digital Document Service Portal

Problem

Academic institutions process document requests — certificates, verifications, transcripts — through manual, paper-based workflows. Requests get lost, status is invisible to applicants, and there's no audit trail. Built a full ERP system to digitize the entire lifecycle from submission to issuance. Won 1st Prize at the college level competition.

Backend Architecture

  • Role-Based Access Control (RBAC): Four distinct roles — Applicant, Clerk, Verifier, Admin — each with scoped API access enforced via JWT middleware. Clerks update status; Verifiers approve; Admins manage users. No role can exceed its boundary.
  • Document State Machine: Each request transitions through defined states: SUBMITTED → UNDER_REVIEW → VERIFIED → APPROVED → ISSUED / REJECTED. Invalid transitions are rejected at the service layer, preventing data corruption from malformed API calls.
  • Audit Logging: Every state change writes to an audit collection with timestamp, actor ID, previous state, and new state — a complete, queryable trail for compliance and dispute resolution.
  • Admin Dashboard API: Filterable endpoints by status and date range. Eliminated 100% of manual document processing for participating departments.
Node.js Express.js MongoDB JWT Auth REST API React
View Source
02

SAAR — AI Voice Assistant

Problem

Built a local AI personal assistant that responds to voice commands, executes system workflows via WhatsApp, and automatically locks the screen when the user leaves — no cloud dependency, no manual interaction required.

System Architecture

  • STT → Wake Word → TTS Pipeline: Custom speech-to-text processing detects a wake word, routes the command to an intent classifier, executes the action, and responds via text-to-speech — all locally, under 2 seconds end-to-end.
  • WhatsApp Automation via Twilio: Commands sent over WhatsApp trigger backend workflow execution — open apps, run scripts, send reminders — making the assistant accessible from any device remotely.
  • OpenCV Presence Detection: Camera feed analyzed in real-time to detect user presence. Sustained absence triggers automatic system lock — security without any user friction.
Python FastAPI OpenCV Twilio STT / TTS Git
View Source
03

PRAVAS — Smart Travel Planner API

Problem

Travel planning involves juggling maps, weather, transit, and accommodation across multiple disconnected apps. Built a backend that integrates real-time APIs to generate a complete, optimized travel itinerary from a single request.

Backend Architecture

  • Multi-API Orchestration: Single endpoint fans out to Google Maps (routing), OpenWeather (forecast), and transport APIs concurrently using async Python — response aggregated and normalized before returning to the client.
  • Route Optimization: Waypoints ordered by proximity and travel time using the Maps Distance Matrix API, minimizing total journey duration for multi-stop itineraries.
  • Dynamic Itinerary Generation: Weather-aware scheduling — outdoor activities deprioritized on rain forecast. Output is a structured JSON plan consumable by any frontend or mobile app.
Python FastAPI Google Maps API OpenWeather API Async / aiohttp REST API
View Source
// skills

Technical Toolkit

◆

Backend

  • Node.js + Express.js REST APIs, middleware, routing, async patterns
  • Python + FastAPI High-performance APIs, async endpoints, AI pipeline integration
  • REST API Design Resource modeling, status codes, versioning, error contracts
  • JWT Authentication Token-based auth, refresh rotation, RBAC middleware
◆

Data & AI

  • MongoDB Schema design, aggregation pipelines, indexing strategies
  • OpenCV Real-time video processing, presence detection, frame analysis
  • STT / TTS Pipelines Speech-to-text, wake word detection, voice response systems
  • Twilio WhatsApp automation, messaging APIs, webhook handling
◆

Tools & Practices

  • Git + GitHub Branching, PRs, commit hygiene, open-source workflows
  • React Frontend for MERN stack projects and admin dashboards
  • Postman API testing, collection runners, environment configs
  • Docker (learning) Containerization, docker-compose for local dev environments
// contact

Let's Connect

I'm looking for a 2-month paid backend or MERN stack internship — remote or India-based. If you're building something that needs solid backend architecture, let's talk.

Email sagarkharbikar25@gmail.com
GitHub github.com/sagarkharbikar25
LinkedIn linkedin.com/in/sagarkharbikar
Phone +91 8208389305