NimbusNerd

Production AI Projects

I design and build applications that integrate advanced cloud engineering with production artificial intelligence. Here are the live, serverless applications I built from scratch using GCP, Gemini, and Edge runtimes.

BudgetYoish

Intelligent Personal Finance Assistant

BudgetYoish reimagines personal finance. Instead of forcing users to scroll through endless spreadsheets or manually categorize transactions, BudgetYoish acts as a real-time financial advisor. It ingests transaction data and delivers immediate, natural language insights about where your money is going and how to save better.

GCPVertex AIGemini 1.5 FlashVercel ServerlessNext.jsTypeScript
Architecture Specs
  • Frontend hosted on Vercel utilizing Next.js App Router.
  • Secure Vercel Serverless functions act as proxy endpoints to keep API keys hidden.
  • Integrates Google Cloud Platform (GCP) Vertex AI API to execute predictions.
  • Uses Gemini 1.5 Flash with structured JSON output configurations to guarantee predictable frontend rendering.
My Role:Creator & Infrastructure Architect

BeJustMe

AI Personality Match & Comparison Engine

BeJustMe is a modern, deep-compatibility matching platform. It bypasses simple swiping mechanics in favor of psychometric modeling. By evaluating user responses, behavioral profiles, and personal manifestos, BeJustMe uses deep generative reasoning to calculate precise compatibility ratings and provides multi-sentence explanations of matches.

SupabaseSupabase Edge FunctionsDenoGCPGemini 1.5 ProTypeScript
Architecture Specs
  • Supabase Database stores profile models and matches.
  • Supabase Edge Functions (Deno TS) handle compute at the edge to secure Vertex AI calls with zero cold-starts.
  • Utilizes GCP IAM Service Accounts and REST OAuth token creation inside Edge Functions.
  • Leverages Gemini 1.5 Pro's high reasoning capability to perform complex comparisons on qualitative behavioral data.
My Role:Creator & Backend Engineer
The Secret Sauce: MLOps on Google Cloud

Both applications represent true cloud-first designs. By removing standard API keys and integrating direct GCP Service Accounts, we gain robust IAM protection. Furthermore, using **Vertex AI** guarantees that our prompts and user transactional data are completely private, compliant, and never used to train Google's public baseline models.