Projects

Production apps across full-stack SaaS, AI infrastructure, and desktop tooling.

Live

Full-Stack Link-in-Bio SaaS Platform

Next.js 16, React 19, TypeScript, Prisma, PostgreSQL, NextAuth v5

Built a production-ready link in bio using Next.js App Router, Server Components, and Server Actions.
Authentication with Google OAuth, credentials login, JWT sessions, email verification, and role-based access.
Designed scalable Prisma/PostgreSQL schemas for profiles, analytics, links, and click tracking systems.
Developed analytics pipelines using request headers, UAParser, and IP geolocation for regional insights.
Integrated UploadThing, React Email, drag-and-drop sorting, interactive charts, and dynamic SEO metadata.

Live

Multi-Tenant AI Support Decision Engine

TypeScript, Node.js, Express, Supabase, pgvector, OpenAI, MCP

Built a multi-tenant AI decision engine enabling workspaces to define skills, triggers, escalation rules, and automated customer support workflows through scalable REST APIs.
Designed and optimized a hybrid retrieval pipeline using OpenAI embeddings, pgvector similarity search, Jaccard overlap ranking, and Reciprocal Rank Fusion (RRF) scoring.
Implemented structured AI decision selection with confidence scoring, escalation handling, and complete decision trace logging for debugging, analytics, and workflow analysis.
Developed secure multi-tenant backend infrastructure with hashed API keys, protected routes, rate limiting, Helmet, CORS policies, and correlation-based request tracing.
Exposed REST and MCP (JSON-RPC) interfaces with Python SDK support, enabling seamless AI agent orchestration and external backend service integrations.

Live

AI Screenshot & Visual Intelligence Utility

Electron, JavaScript, Groq/OpenAI APIs, IPC, Chromium Capture APIs

Built a cross-platform Electron tray app for AI screenshot capture, OCR, and visual analysis workflows.
Developed fullscreen region capture using desktopCapturer, desktop streams, and canvas-based crop selection.
Designed a secure Electron architecture with preload contextBridge APIs and isolated renderer processes.
Integrated Groq and OpenAI multimodal APIs to power screenshot-based AI chat, OCR text extraction, contextual image understanding, and advanced visual reasoning workflows.
Implemented stealth/privacy mode, encrypted credentials, and installers for Windows, Linux, and macOS.