Yevhenii Melnyk
Offer Yevhenii work on your next project.
Private message
Offer work
available for hire
Rating
Successful projects
Average rating
Skills and abilities
Portfolio
-
Lyubava AI Bot
Bot Development
Lyubava AI — Empathetic Digital Avatar
Development of a high-load AI avatar with integration of the Telegram Stars payment system and an adaptive role model based on Llama 3.
Stack: Python (Aiogram 3), Groq API (Llama 3.3 70B), ElevenLabs API, SQLite (aiosqlite), Telegram Stars.
…
Concept
Lyubava is not just a chatbot, but a complex AI role model that imitates the behavior of a real person. The main goal of the project: to create a high level of retention through emotional attachment and deep personalization.
Key Technical Features
1. Multi-level memory and entity extraction system
Automatic parsing: The bot does not just rewrite, it "listens." In the background, Llama-3 operates, extracting the user's name, age, and gender from the context without direct questionnaires.
Long-term memory: Using SQLite to store dialogue history and facts about the user, allowing the bot to recall details even weeks after conversations.
2. Dynamic style adaptation (Adaptive Persona)
The bot changes the communication archetype depending on the user's age: from "Older Sister" for teenagers to "Experienced Woman" for mature men.
Implemented logic of "psychological boundaries" (inaccessibility), making the AI less like an algorithm and more like a personality with character.
3. Monetization through Telegram Stars (XTR)
Integration with the official Telegram payment system for selling digital goods.
Private Mode: Gamified access to exclusive content (voice messages, flirt mode).
Voice Limits: Optimization of costs on ElevenLabs API through a limit system in the database, making the project economically sustainable.
4. Multimodality
Voice: Realistic speech synthesis through ElevenLabs with customized "humanity" parameters (sighs, intonations).
Video: A system for sending video messages ("circles") and reactions to reinforce positive experiences (gifts/donations).
Business Metrics (Value Proposition)
Retention: Psychological flirting and "hard-to-get" increase user excitement and return frequency.
Scalability: The code is built on asynchronicity (Asyncio/Aiogram), allowing it to handle thousands of simultaneous dialogues.
Low Latency: The use of Groq API ensures virtually instantaneous AI response (up to 300 tokens per second), which is critical for simulating a "live" chat.
Activity
| Latest proposals 1 | Budget | Added | Deadlines | Proposal | |
|---|---|---|---|---|---|
|
Create an AI agent for communication
180 USD
|
