React + TypeScript
Guided campaign flows, rich-message editors, chat workspaces, reports, bots, and localized settings.
← Back to projects
02 / Featured product
TGBotSender
A local-first Telegram workspace for account and bot operations, targeted campaigns, rich messages, conversations, reporting, and bring-your-own-key AI assistance.
ProductionWindowsmacOSLocal-first
/overview/product.md
One Telegram workspace.
TGBotSender unifies connected Telegram accounts, contacts, groups, chats, campaigns, saved audiences, templates, auto-reply rules, AI agents, bot inboxes, and reports in a single desktop product.
My work covers product planning, system architecture, React and TypeScript UI, Tauri and Rust native development, SQLite persistence, Telegram integrations, AI provider abstraction, secure credential handling, packaging, localization, and release maintenance.
/architecture/system.map
Provider-aware design.
The application separates interface state, native commands, durable local records, and external providers. That boundary supports multiple Telegram senders and AI providers without placing sensitive tokens in ordinary UI state.
Tauri IPC
A narrow command surface connects desktop UI actions to native services and long-running tasks.
Rust services
Telegram operations, queues, rate-limit handling, credentials, licensing, and background execution.
SQLite + APIs
Local operational state connects to Telegram and the user's selected AI provider when requested.
/engineering/decisions.log
Controlled automation.
The core challenge is coordinating many long-running, provider-limited workflows while keeping progress, failures, and user control visible.
Campaign state
Audience, content, personalization, timing, pause, resume, cancel, history, and recipient outcomes stay attached to a durable job.
Rich-message tooling
Visual editors cover formatting, media, polls, locations, inline buttons, reply keyboards, and sender-aware capability rules.
Unified conversations
Accounts, bots, inboxes, chats, contacts, groups, and human handoff are organized without switching products.
BYOK AI
OpenAI, Gemini, Anthropic, and OpenRouter are implemented behind a provider abstraction with model settings and reusable actions.
Secure credentials
API keys and account secrets use operating-system credential storage, while application data remains in local persistence.
Failure handling
Queues, retries, eligibility checks, cooldowns, human approval modes, and exact Telegram rate-limit states keep automation inspectable.
/next/connect.sh
Building a product with complex workflows?
I turn integration-heavy requirements into structured interfaces, durable jobs, secure provider boundaries, and maintainable cross-platform releases.