Ultra-Fast MTProto Telegram Bot
Budget: ₹20,000 – ₹50,000 INR
I need a Telegram user-level bot that watches one specific chat (or a message ID I give it) and fires off the callback for every inline button the instant it appears. The emphasis is raw speed and consistently beating the latency I get from standard Telethon / Pyrogram implementations.
Because I want the absolute minimum overhead, please work directly with the MTProto layer (TDLib, raw API, or your own lightweight client). The language itself is secondary; what matters is that the implementation uses MTProto calls rather than high-level wrapper logic, so feel free to reach for C++, Python, or even mix in Rust if it helps keep the event loop razor-thin.
Core expectations
• Handles both inline buttons and their callback queries in a single flow, with aggressive parallel parsing so the answerCallbackQuery call goes out immediately.
• Internal architecture designed around async I/O: a tight event loop, minimal context-switching, and non-blocking network operations.
• Built-in speed test: a script or harness that spawns a local echo bot (or similar) and prints the average and 95th-percentile latency from button appearance to callback confirmation so I can verify your numbers.
• Safety valve for Telegram rate limits—no bans, even under quick bursts—without lengthening normal latency.
• Clean source, README, and deployment steps for both Linux VPS and a local test environment.
Acceptance criteria
1. End-to-end latency under 120 ms on a mid-range VPS in singapore (measured by the included speed-test script).
2. No missed button presses in a 1,000-message test run.
3. Code compiles/runs with a single command and uses only publicly documented Telegram APIs.
If you can squeeze out more speed or have benchmarking ideas I haven’t considered, call them out—raw performance wins the bid.
Because I want the absolute minimum overhead, please work directly with the MTProto layer (TDLib, raw API, or your own lightweight client). The language itself is secondary; what matters is that the implementation uses MTProto calls rather than high-level wrapper logic, so feel free to reach for C++, Python, or even mix in Rust if it helps keep the event loop razor-thin.
Core expectations
• Handles both inline buttons and their callback queries in a single flow, with aggressive parallel parsing so the answerCallbackQuery call goes out immediately.
• Internal architecture designed around async I/O: a tight event loop, minimal context-switching, and non-blocking network operations.
• Built-in speed test: a script or harness that spawns a local echo bot (or similar) and prints the average and 95th-percentile latency from button appearance to callback confirmation so I can verify your numbers.
• Safety valve for Telegram rate limits—no bans, even under quick bursts—without lengthening normal latency.
• Clean source, README, and deployment steps for both Linux VPS and a local test environment.
Acceptance criteria
1. End-to-end latency under 120 ms on a mid-range VPS in singapore (measured by the included speed-test script).
2. No missed button presses in a 1,000-message test run.
3. Code compiles/runs with a single command and uses only publicly documented Telegram APIs.
If you can squeeze out more speed or have benchmarking ideas I haven’t considered, call them out—raw performance wins the bid.