Pawan Choudhary
I build AI-powered applications, modern web experiences, and developer tools with a focus on thoughtful design, performance, and real-world impact.
Pawan Choudhary
AI Developer • Full Stack Engineer
Mindset
Discipline, consistency, and curiosity shape how I approach both engineering and life. Improving my body, strengthening my mind, and sharpening my technical skills all feed into becoming a better problem solver.
Craft
I care deeply about software that feels polished from every angle — clean architecture, accessible interfaces, and performance that holds up under real use.
India
- Location
- East Delhi, New Delhi
- PIN Code
- 110092
- Timezone
- IST · UTC +05:30
My Toolkit
A working set of the languages, frameworks, and tools reached for daily — the full categorized breakdown lives on the dedicated Skills page.
Curated Work
Turns a resume PDF into a deployable portfolio site, analyzed by a local AI model, not a third-party API
AI Portfolio Generator
A full-stack portfolio generator that extracts a resume's content and runs it through a locally-hosted Qwen3 model via Ollama, then renders the result into one of five real themes; live in the browser before anything is exported as a self-contained HTML file or a zipped static site. Nothing about the pipeline is a thin wrapper around an API call: extraction, analysis, rendering, and export are each their own stage, with the model running on infrastructure I control rather than someone else's endpoint.
- Resume PDF upload with structured extraction of experience, projects, and skills
- Local AI analysis via Ollama + Qwen3, nothing leaves the machine, no third-party API calls
- Five distinct themes (Minimal, Executive, Developer, Creative, Modern SaaS), different layouts, not palette swaps
- Instant theme switching that only re-renders the page, without re-running extraction or AI analysis
- Live in-browser preview before exporting a self-contained HTML file or a zipped static site
AI Portfolio Generator
A full-stack portfolio generator that extracts a resume's content and runs it through a locally-hosted Qwen3 model via Ollama, then renders the result into one of five real themes; live in the browser before anything is exported as a self-contained HTML file or a zipped static site. Nothing about the pipeline is a thin wrapper around an API call: extraction, analysis, rendering, and export are each their own stage, with the model running on infrastructure I control rather than someone else's endpoint.
- Resume PDF upload with structured extraction of experience, projects, and skills
- Local AI analysis via Ollama + Qwen3, nothing leaves the machine, no third-party API calls
- Five distinct themes (Minimal, Executive, Developer, Creative, Modern SaaS), different layouts, not palette swaps
- Instant theme switching that only re-renders the page, without re-running extraction or AI analysis
- Live in-browser preview before exporting a self-contained HTML file or a zipped static site
An explainable decision engine for weather-driven decisions, not just another weather display
Weather Dashboard
Ingests raw meteorological data from a keyless provider and runs it through a rules-based recommendation engine that answers specific, practical questions (should I carry an umbrella, is today good for a run, is the air safe to breathe, will this disrupt travel) with a plain-English reasoning trace behind every verdict, and a documented fallback path if the primary provider goes down.
- Five independent recommendation rules: umbrella, running suitability, outdoor activity, travel suitability, and air quality
- Every verdict ships with a reasoning string citing the exact factor that drove it, not just a score
- Provider fallback: Open-Meteo (keyless) primary, OpenWeatherMap fallback, resolved behind one orchestration point
- Per-data-type cache TTLs instead of one blanket caching policy
- Responsive dashboard that collapses cleanly from a desktop grid to a single mobile column
Weather Dashboard
Ingests raw meteorological data from a keyless provider and runs it through a rules-based recommendation engine that answers specific, practical questions (should I carry an umbrella, is today good for a run, is the air safe to breathe, will this disrupt travel) with a plain-English reasoning trace behind every verdict, and a documented fallback path if the primary provider goes down.
- Five independent recommendation rules: umbrella, running suitability, outdoor activity, travel suitability, and air quality
- Every verdict ships with a reasoning string citing the exact factor that drove it, not just a score
- Provider fallback: Open-Meteo (keyless) primary, OpenWeatherMap fallback, resolved behind one orchestration point
- Per-data-type cache TTLs instead of one blanket caching policy
- Responsive dashboard that collapses cleanly from a desktop grid to a single mobile column
A privacy-focused music streaming app built so the essential features are never locked behind a paywall
Armonia
Armonia is a privacy-focused music streaming app for Android, built on Flutter with Supabase as the backend and Google Sign-In for authentication. Guests get a fully local experience with no account required, and signed-in users have their data synchronized through Supabase instead of trusting it to local storage alone. The whole app is built around one idea: music should be accessible to everyone, without the essential features held back behind a subscription.
- Privacy-focused streaming with a fully local guest mode, no account required to use the app
- Google Sign-In for authenticated users, with data synchronized through Supabase instead of local storage alone
- Clean, simple navigation designed to scale with future features without becoming cluttered
- Core features available without a subscription, rather than paywalling basic functionality
Armonia
Armonia is a privacy-focused music streaming app for Android, built on Flutter with Supabase as the backend and Google Sign-In for authentication. Guests get a fully local experience with no account required, and signed-in users have their data synchronized through Supabase instead of trusting it to local storage alone. The whole app is built around one idea: music should be accessible to everyone, without the essential features held back behind a subscription.
- Privacy-focused streaming with a fully local guest mode, no account required to use the app
- Google Sign-In for authenticated users, with data synchronized through Supabase instead of local storage alone
- Clean, simple navigation designed to scale with future features without becoming cluttered
- Core features available without a subscription, rather than paywalling basic functionality
Thoughts & Writings
My Workflow
A glimpse at the everyday software behind the work — the full setup, hardware included, lives on the dedicated Uses page.