A personal AI assistant in the form of a Telegram Mini App for automating nutrition control and dietary analysis.
What the app does:
AI Scanner (Gemini API in this case): Recognizes products via photos or text descriptions. No more searching for every ingredient in a database — the neural network automatically determines weight, calories, and nutrients.
Personal Calculation: A multi-step onboarding process that calculates individual calorie and macro targets based on the user's metabolism and goals.
Interactive Dashboard: Visual daily progress monitoring through activity rings and dynamic macronutrient cards.
Telegram Native: Operates as a full-fledged application within the messenger (Mini App). Fast, one-click access without downloading third-party software.
Tech Stack:
Frontend: React 19, TypeScript.
Backend & AI: Bun runtime, Google Gemini API, Prisma ORM.
Database: PostgreSQL.
UI/UX: Tailwind CSS, Shadcn UI (Mobile-first design).
Engineering Features:
Edge-ready Architecture: High responsiveness (Latency) due to the Bun environment and optimized server requests.
Server Components (RSC): Offloading heavy computations to the server side, ensuring lightning-fast performance on any smartphone.
Type-safe Data Layer: End-to-end typing of all data — from neural network responses to the DB schema, eliminating application errors.
Advanced Prompt Engineering: Refined AI interaction algorithms to achieve maximum calculation accuracy without "hallucinations."
What the app does:
AI Scanner (Gemini API in this case): Recognizes products via photos or text descriptions. No more searching for every ingredient in a database — the neural network automatically determines weight, calories, and nutrients.
Personal Calculation: A multi-step onboarding process that calculates individual calorie and macro targets based on the user's metabolism and goals.
Interactive Dashboard: Visual daily progress monitoring through activity rings and dynamic macronutrient cards.
Telegram Native: Operates as a full-fledged application within the messenger (Mini App). Fast, one-click access without downloading third-party software.
Tech Stack:
Frontend: React 19, TypeScript.
Backend & AI: Bun runtime, Google Gemini API, Prisma ORM.
Database: PostgreSQL.
UI/UX: Tailwind CSS, Shadcn UI (Mobile-first design).
Engineering Features:
Edge-ready Architecture: High responsiveness (Latency) due to the Bun environment and optimized server requests.
Server Components (RSC): Offloading heavy computations to the server side, ensuring lightning-fast performance on any smartphone.
Type-safe Data Layer: End-to-end typing of all data — from neural network responses to the DB schema, eliminating application errors.
Advanced Prompt Engineering: Refined AI interaction algorithms to achieve maximum calculation accuracy without "hallucinations."