Hi, I'm Shivaraj
Full-stack developer shipping production web and mobile products with React Native, Next.js, and TypeScript.
SK

About Me

Full-stack developer with 2 years of hands-on experience shipping production web and mobile products using React Native, Next.js, TypeScript, Node.js, and Supabase/PostgreSQL. At Vokido, I built WhatsApp growth loops, payments, and reliability features for a voice-first AI tutor. Previously built role-based asset management platforms adopted by 3 organizations and sports performance tracking dashboards for 200+ users. Currently pursuing my B.E. in AI & Data Science (2026).

Experience

V

Vokido

July 2026 - Present
Member of Technical Staff
Shipping growth, payments, and reliability for a voice-first AI English tutor app like WhatsApp report cards, viral badge sharing, attribution, and Play Store release automation.
  • ●Built the parent engagement loop with automated, tiered WhatsApp report cards rendered with Satori + resvg in Supabase Edge Functions, shipped behind a kill switch with a 10% holdout and Sentry drop alerts, achieving 100% delivery rate; fixed silent failures at Edge's CPU limit by profiling rasterization (764ms to 282ms), capping renders per tick, and vendoring WASM + fonts for zero-network rendering.
  • ●Created a viral share loop with personalised badge images (child's name on artwork) rendered to PNG, cached in Supabase Storage, and sent as Meta WhatsApp template messages with 100% delivery rate without blocking session end; also shipped an automated trial-end WhatsApp campaign (day 0/3/8) with a campaign ledger, send crons, poster creatives, and STOP-intent handling to recover expiring trials.
  • ●Designed deep-link and install attribution across backend, Android and web (resolve/click APIs, IP throttling, verified App Links), plus UTM tracking that gave partner signups measurable sources.
  • ●Integrated Fastlane with the existing GitHub Actions pipeline to automate Play Store releases, reducing manual deploy time; validated end to end against Google Play.
  • ●Rebuilt the paywall with offer-first plan selection, intro-offer date math across review/success screens, and server-driven limits so ops can retune allowances without an app release.
  • ●Made voice chats survive upstream outages with retry-with-backoff, machine-readable 503s, and a degrade-to-text fallback so a TTS failure downgrades one reply instead of killing the session.
S

Sportality Pvt Ltd

May 2025 - February 2026
Product Developer Intern
Built an athlete management platform with role-based access and performance dashboards serving 200+ active users.
  • ●Built an athlete management platform with role-based access for admins, coaches, and athletes serving over 200 active users.
  • ●Created dashboards showing athlete performance metrics and training progress to help coaches track team development, translating their feedback into features that made daily team monitoring more intuitive.
  • ●Added server-side pagination and API caching which cut response times by 40% during peak usage hours.
  • ●Worked on modular architecture separating business logic from UI components for easier maintenance and testing.
V

Vulnuris

September 2024 - May 2025
Junior Product Developer Intern
Built an ITSM platform adopted by 3 organizations with asset tracking, SBOM, SLA monitoring, and secure RBAC.
  • ●Built ITSM platform adopted by 3 organizations with asset tracking, SBOM management, SLA monitoring, and ticketing system.
  • ●Created dashboards using Chart.js and Recharts to help IT teams visualize infrastructure health and resource usage, and added role-based access control and two-factor authentication using JWT to secure sensitive operations.
  • ●Designed the ticketing and SLA features around IT teams' asset management workflows, and built automated workflows for license renewals and SBOM updates that reduced manual tracking effort by 50%.
  • ●Optimized PostgreSQL queries with proper indexing and used React context for state management, making the UI 30% faster.

Tech Stack

languages

JavaScript
TypeScript
Python
C++

frontend

React
React Native
Expo
Next.js
TailwindCSS
Shadcn UI
Zustand
HTML
CSS

backend

Node.js
Express.js
Elysia.js
FastAPI
Deno
JSON Web Token
Drizzle ORM
Better-auth

databases

PostgreSQL
MongoDB
MySQL
Redis
SQLite

tools

Docker
Amazon Web Services
GitHub Actions
Fastlane
Grafana
Prometheus
Loki
Supabase
Appwrite
Upstash
Nginx
Vercel
Netlify
Git
GitHub
Figma
PROJECTS

Things I've Built

A collection of projects I've worked on, ranging from AI-powered applications to scalable web platforms. Each one taught me something new.

Better Chat

Better Chat

Real-time chat with WebSocket messaging, SSE notifications, Web Push alerts, friends, and group conversations with join requests. Includes an AI agent system of 6 agents on Groq and the Vercel AI SDK that send email, manage Google Calendar/Docs, and create TickTick tasks with Upstash QStash background jobs, Redis and rate limiting in a Turborepo + Bun monorepo.

Next.js
Elysia.js
TypeScript
PostgreSQL
Drizzle
Upstash Redis
Upstash QStash
Better-Auth
Vercel AI SDK
SparkChat

SparkChat

AI chat application with a ChatGPT-like interface, suggested prompts, and organized conversation threads. Supports multiple models including Google Gemini and Meta Llama with streaming responses, Clerk authentication, PostgreSQL chat history, and rate limiting for handling 100+ concurrent users with sub-second response times.

Next.js
TypeScript
Neon PostgreSQL
Drizzle ORM
Vercel AI SDK
Clerk
Burner Rooms

Burner Rooms

A privacy-first, lightweight real-time chat application designed for ephemeral conversations. Enables users to create temporary chat rooms with custom time limits and participant limits. Features client-side anonymous identity generation, real-time messaging powered by Upstash Realtime with emoji and thread support, automatic room expiration with TTL, and minimal data retention.

Next.js
Elysia
TypeScript
Upstash Redis
Upstash Realtime
TanStack Query
TailwindCSS
shadcn
Learn Orbit

Learn Orbit

An interactive learning platform designed for CS students to master new skills through structured roadmaps, curated resources, and hands-on projects. Combines guided learning paths with practical application.

React
TailwindCSS
Shadcn UI
Clerk
Vercel
EasyLearn AI

EasyLearn AI

An intelligent study companion powered by Generative AI. Transform PDFs into interactive learning materials with AI-generated notes, smart flashcards, practice quizzes, and context-aware chat. Includes mock interview features for comprehensive exam preparation.

Python
FastAPI
Google Generative AI
React
TailwindCSS
HACKATHONS

Building Under Pressure

Participating in hackathons has been an incredible way to push boundaries, learn new technologies, and collaborate with amazing people. Here are some highlights from my journey.

  • C

    Code Cubicle Hackathon

    Developed an intelligent stock market AI agent with multi-modal interaction capabilities. Users can query real-time stock information through both text and voice inputs, enabling hands-free access to market data, price trends, and financial insights. The agent processes natural language queries and delivers comprehensive stock analysis.

  • T

    The Better Hack Hackathon

    Top 5

    Created an AI-powered music production tool designed as a cursor extension for music producers. The Electron-based desktop application enables users to generate music beats using AI, streamlining the creative workflow. Features real-time beat generation, customizable parameters, and seamless integration with popular DAWs.

  • S

    Student Hacks

    Winner

    Built an AI-powered interview preparation platform that conducts realistic mock interviews. The system uses advanced NLP to analyze responses, providing comprehensive feedback on communication skills, technical knowledge, and interview performance. Generates detailed reports highlighting strengths, areas for improvement, and personalized recommendations for interview success.

  • F

    Fetch AI

    Top 5

    Developed a machine learning-powered sales forecasting system that processes Excel data files to predict future sales trends. The model analyzes historical sales data, identifies patterns, and generates accurate 3-month sales predictions. Features include data preprocessing, ML algorithm integration, and interactive visualization of forecasted results.

LET'S CONNECT

Let's Build Something Together

I'm always interested in hearing about new opportunities, interesting projects, or just connecting with fellow developers. Feel free to reach out on LinkedIn or check out my code on GitHub.