Intro

Anshul Bansal

Principal Engineer, Web Platform · OLX India · since Apr 2026 · Gurugram, India

I architect the web platform for OLX India’s marketplace — PWA architecture, SSR for panamera-web / www.olx.in, Akamai and CDN delivery, BFF patterns, Core Web Vitals, observability, SEO, and client-side AI — across about 35M users. I lead three PWA squads on that surface.

Eleven years in, eight of them at OLX: I joined as Web Developer (SE1) in 2018 and moved through SE2, SE3, Technical Architect / Chapter Lead, into Principal. Before that I built product UI and APIs at FabHotels and Homelane, starting in 2015.

Experience

  1. – Present

    OLX India

    Principal Engineer

    • Driving lightweight SSR for bot traffic to improve crawl efficiency and reduce bandwidth cost
    • Optimising Listing-page LCP (mobile) toward the <2.5s threshold; advancing INP on interaction-heavy pages
    • Scaling Akamai cache hit ratio beyond 60%; setting technical direction for BFF adoption across web and Android
  2. –

    OLX India

    Technical Architect · Chapter Lead (Web)

    • ~50% monthly infra cost reduction at the same traffic scale; CDN cache hit ~30% → ~55–60%; 80% pod-count reduction via CPU scaling and Akamai offload
    • 45–50% LCP improvement on Desktop Home; 30–50% INP on interaction-heavy pages; CLS stabilised near zero; led technical SEO to +160% daily organic clicks and +50% impressions over 14 months
    • 80–99% fewer 5xxs on high-traffic APIs; ~50K abusive bot / parasite-SEO requests blocked daily; performance budgets and Web Vitals gates in CI
  3. –

    OLX

    Lead Engineer (SE3)

    • Designed and owned end-to-end architecture for the car-dealer Super App using React and Server-Driven UI
    • Built and scaled the advertisement platform (React, Node.js, DynamoDB); mentored a cross-functional engineering team
  4. –

    OLX

    Senior Software Engineer (SE2)

    • Developed the car-dealer Super App with React and Server-Driven UI for high-volume dealer workflows
    • Built and maintained the advertisement platform supporting OLX India’s core monetisation surface
  5. –

    OLX

    Web Developer (SE1)

    • Enabled microfrontends via Web Components in the PWA; shipped PWA features with React–Redux
    • Built B2C tools and the monetisation admin platform with Vue.js and React–Redux
  6. –

    FabHotels

    Frontend Developer

    • Developed new features for the FabHotels mobile website and set up the CI/CD pipeline for mobile-web deployments
  7. –

    HomeLane

    Software Engineer

    • Built the company site and a responsive microsite in React–Redux; developed the internal CRM on the MEAN stack
    • Wrote Node.js REST APIs for escalations, AWS document upload, and cross-platform data sync

Work

Platform work at OLX — selected threads from recent years.

Edge caching and bot classification

Traffic classification and differential caching across page types so humans, search crawlers, and other bots hit the edge strategy that fits them. CDN cache hit moved from ~30% to ~55–60%; about 50K abusive bot and parasite-SEO requests are blocked daily, protecting crawl quality and origin load.

SEO at scale

Technical and on-page SEO work over 14 months drove about +160% daily organic clicks and +50% impressions, with a 99.5% reduction in 404s via expired-listing redirects and zero-inventory suppression — written up in two parts on tech.olx.in.

AI discoverability

Making the site legible to LLM crawlers: robots.txt Content-Signal directives, an llms.txt, and markdown content negotiation so agents can pull structured answers without scraping the full PWA shell.

Googlebot event-loop incident, Apr 2026

After a core update, deep filter and tracking-param URLs saturated the Node event loop. HPA was scaling on the wrong signal; vertical CPU scaling restored headroom while the URL surface was tightened.

Trunk-based development with session flags

Session-scoped flags in the SSR pipeline let the team merge small changes continuously without exposing unfinished work. Merge time halved and PR throughput rose about 4× — written up on tech.olx.in.

Projects

Health Report AI

Local-first lab-report digitizer and dashboard: scanned PDFs and images go through a vision model on Ollama (Qwen2.5-VL), land as structured JSON, and surface in a React + Flask UI for trends, search, and AI analysis. No report data leaves the machine.

ReviewOwl

GitLab CI + Slack bot for automated MR reviewer assignment with load-aware round-robin. In daily use by my team.

OLX Assist

Workflow-based chat widget for help.olx.in. JS SDK embed, cookie/Bearer auth, and XMPP-based proactive triggers.

Earlier open source

  • Gourmet-Food — Food ordering app using Zomato API
  • Chatter — Real-time chat app with Socket.io
  • TaskR — Trello-like task manager application
  • PUBG-Rank-Predictor — Machine Learning model to predict PUBG player rankings
  • Human-Activity-Tracker — ML model to classify human activities from sensor data
  • React-Card-Layout — Responsive card layout component library
  • Lazy-Loader — Performance optimization library for lazy loading images
  • UrbanMart — Modern e-commerce platform with product catalog, shopping cart, and responsive design

Writing

Speaking

Delivered

Your Server-Rendered App Doesn't Know Who It's Talking To

  • Dev Days Mumbai ·
  • Tech Talks @ OLX India ·

Modularising OLX's frontend monolith with Web Components

  • Naspers Global Tech Conference, Amsterdam ·

Upcoming

Recorded talks

Microfrontends and Web Components · Predicting the winner in a game of PUBG · Globalisation Of OLX Code Base

Teaching

I mentor engineers via Topmate and internal workshops, ran a clean-code workshop for the intern bootcamp, and organised OLX India Hackathon 2026 (~80 participants).

I also speak on leadership and communication: Identifying Your Leadership Style, Ice Breaker Speech, Let's Team Up.

Skills

Platform

PWA architecture, SSR at scale, BFF patterns, micro-frontends, Web Components, Server-Driven UI

Edge & performance

Akamai, CDN strategy, caching, Core Web Vitals, SEO

Observability

New Relic, NRQL, incident response, capacity planning

Everything else

JavaScript, TypeScript, Node.js, React, Svelte, Python, GenAI / local vision models (Ollama)

Background

B.Tech in Computer Science

USICT, GGSIPU · Delhi · 2011 – 2015 · 77%

Comprehensive computer science education covering programming fundamentals, data structures, algorithms, software engineering, and system design.

Certifications & awards

Contact: anshulbansal2209@gmail.com