Remindr
A context-aware personal assistant that helps users follow through on tasks, deadlines, and goals by combining planning logic, connected context, and proactive delivery.
Remindr is designed around a stronger premise than traditional reminder apps: timing is part of the problem, not just the task itself. Instead of relying on static reminders, it combines task metadata, calendar context, fatigue patterns, and planning signals to decide when a nudge should happen.
The system includes a web control surface, backend APIs, background workers, external connectors, and Telegram delivery. That gives the product a broader scope than a simple to-do manager and makes it feel closer to a proactive assistant than a passive tracker.
Inspiration Behind It
The project comes from the gap between what users plan to do and when they are actually in a position to act. Static reminders often fail because they assume the user already knows the right timing, even when energy, schedule pressure, and context keep shifting.
Remindr treats follow-through as a context problem. It is built to notice urgency, fatigue, deadlines, and routine signals so the product can surface actions when they are more likely to be useful instead of becoming background noise.
System Architecture
Architecture Diagram
Remindr
10 nodes
About The Flow / Architecture
Remindr is split into a React frontend, a FastAPI backend, a Postgres or Supabase-backed data layer, and background workers for sync and notification workflows. The frontend handles onboarding, integrations, and dashboard surfaces, while the backend exposes routes for tasks, connectors, fatigue, notifications, decisions, and Telegram delivery.
A key part of the system is the decision and service layer, which combines user preferences, deadlines, internal calendar blocks, and fatigue signals into recommendations. Workers and watchers then handle ongoing sync, notification dispatch, memory distillation, and deadline monitoring outside the request cycle.
Remindr combines scheduling, context ingestion, notification logic, and proactive delivery into a single assistant-style workflow. The current codebase already includes onboarding, task management, connector sync, internal planning blocks, fatigue tracking, and Telegram-based communication.
It is a strong portfolio project because it blends backend system design, product thinking, worker orchestration, and assistant behavior into one cohesive platform rather than a simple CRUD application.