Skip to content
PBPrathamesh Basagare
Discuss a project

About

WordPress engineer who builds full-stack, from plugin internals to Next.js and AI.

I design and build custom WordPress plugins, REST APIs and performance-critical front ends, and ship Next.js apps with LLM integrations. The principles carry across stacks: clear data models, slow work moved off the request path, explicit API contracts, and measuring before optimising.

Request my resume GitHub LinkedIn

01 — Capabilities

What I build, and how I think about it.

Custom WordPress engineering

Plugins and themes built from scratch for requirements off-the-shelf plugins can't meet, structured so they survive core and plugin updates.

  • Plugin architecture with hooks, filters and custom tables
  • Background jobs with WP-Cron and Action Scheduler queues
  • Stripe Elements payment integrations
  • Custom Gutenberg blocks and theme development

APIs & front end ↔ back end

The contract between client and server is where most bugs hide. I make it explicit, validated and boring.

  • Custom REST endpoints with permission callbacks and validation
  • Typed request and response handling in TypeScript
  • Error handling, retries and rate-limit awareness
  • Next.js Server Actions and Route Handlers

Performance engineering

Diagnosing why a page is slow, element by element, then fixing the cause rather than stacking another cache plugin on top.

  • Core Web Vitals diagnosis (LCP, CLS, INP)
  • Render-blocking resources, fonts and image pipelines
  • Page, object and CDN caching layers
  • Static generation and edge delivery

AI integration

LLM features that behave like software, not demos: structured output, fallbacks and predictable failure modes.

  • LLM output constrained to a typed JSON component schema
  • Ranked multi-model fallback chains (Gemini, OpenRouter)
  • Graceful handling of quota and rate-limit errors
  • Generative UI rendered from structured output

Debugging & troubleshooting

Finding the real cause of production issues across PHP, JavaScript, the database and the network, and leaving it easier to diagnose next time.

  • Root-cause analysis across the full request path
  • Slow-query investigation and fixes
  • Reproducing environment-specific bugs safely
  • Logging that shortens the next incident

Technical SEO & indexing

Making sure search engines can find, understand and index what was built.

  • Structured data (JSON-LD) and canonical URLs
  • XML sitemaps and robots rules
  • Google Search Console indexing workflows
  • Semantic, accessible HTML

02 — AI-augmented workflow

AI drafts. I design, review and own every line.

I use AI tools like Claude Code to move faster across stacks, not to skip the thinking. The design decisions, the review and the responsibility stay with me.

  1. 1

    Specify

    Write down the requirement, edge cases and acceptance criteria before any code is generated.

  2. 2

    Design

    Choose the data model, boundaries and failure modes myself. AI is a sounding board, not the architect.

  3. 3

    Generate in small slices

    AI drafts implementation and tests one reviewable piece at a time, never a whole feature in one go.

  4. 4

    Review every diff

    Read each change, reject what doesn't fit the architecture, and refactor until it reads like code I wrote by hand.

  5. 5

    Verify and measure

    Type checks, linting, tests and a real browser before shipping, then performance and error checks after release.

03 — Stack

Tools I use in production.

Languages
PHPTypeScriptJavaScriptSQLHTMLCSS
Frameworks
WordPressNext.jsReactTailwind CSS
Back end & data
WordPress REST APIMySQLAction SchedulerWP-CronServer Actions
Integrations
Stripe ElementsGmail APIThird-party REST APIs
AI
Claude CodeGemini APIOpenRouterStructured LLM output
Performance & SEO
LighthouseCore Web VitalsGoogle Search ConsoleJSON-LD
Tooling
GitGitHubVercelKeystaticGutenberg blocks

04 — Open work

Code you can read.

05 — How this site is built

This page is a working sample.

Next.js 16 App Router, TypeScript strict, Tailwind CSS v4. Built to the same performance and SEO standard I deliver to clients.

Rendering
Every public route statically generated at build time — no database, no runtime fetch.
Content
Keystatic git-based CMS: content lives in the repo as Markdoc, edits commit to GitHub and trigger a rebuild.
Code highlighting
Shiki runs at build time, so code blocks ship zero client JavaScript.
Contact form
Server Action → Zod validation → honeypot + Turnstile → rate limit → Gmail API over OAuth2.
Performance
Server Components by default; three small client islands. The footer shows this page's real LCP, CLS and JS weight.
SEO
Canonical URLs, generated OG images, JSON-LD (Person, WebSite, ProfilePage, BlogPosting), sitemap and robots.