Temirlan Kdyrkhan
Professional Profile
Professional Summary
Experienced Machine Learning Engineer with 5+ years of expertise in AI, algorithmic trading, and full-stack development. Proven track record in developing reinforcement learning models, building trading systems, and creating scalable web applications. Strong background in Python, PyTorch, Java, and modern web technologies with a passion for solving complex problems through innovative technology solutions.
Professional Experience
2025 - Current
Algorithmic Trader
• Developed custom MT5 indicators using volatility bands and ATR for finding optimal Take Profit and Stop Loss levels • Created more than 5 custom trading indicators with advanced technical analysis algorithms • Achieved over 1000 downloads on MetaTrader 5 platform, demonstrating market validation and user adoption • Implemented real-time market analysis tools for automated trading strategies and risk management
2023 - 2025
FAU - Graduate Researcher / Student Projects
• Researched fraud detection using DNN and imbalance learning, deploying AI models and web apps with Python, PyTorch, Flask, and GCP • Built multi-stage Docker containers to streamline deployment and reproducibility • Developed an automated Telegram message retrieval for trading signals with NLP and real-time data handling
2022
FOXI Boats - Software Engineer Intern
• Modernized UI by migrating legacy Java code to Kotlin Jetpack Compose, improving usability and visual appeal • Reduced app crash rates by 30% through debugging and landing page optimization • Improved data retrieval by debugging SQL queries, reducing execution time by 10% • Improved data retrieval speed by 25% via schema redesign and Snowflake-based storage solutions
2021
Bank CenterCredit JSC - Software Engineer Intern
• Integrated Alseco bill payment API in microservices, resulting in 20% reduction in transaction processing time • Boosted service efficiency through the development of optimized SQL queries, B2B operations • Achieved a 95% success rate in integration testing for critical services, reducing post-deployment issues
2018 - 2021
Prophi - Math and Computer Science Mentor
• Delivered personalized instruction on the fundamentals of data structures to students worldwide • Mentored students in mathematics and computer science concepts • Developed teaching materials and curriculum for international students
Education & Certifications
2023 - 2025
Florida Atlantic University - Masters in Data Science and Analytics
Currently pursuing Masters in Data Science and Analytics with a perfect 4.0 GPA. Focus on advanced machine learning, deep learning, and AI applications in real-world scenarios.
2017 - 2021
Lomonosov Moscow State University - Bachelor of Mathematics
Earned Bachelor of Mathematics degree with a 3.25 GPA. Strong foundation in mathematical principles, statistical analysis, and computational methods essential for data science and AI.
Technical Skills
Machine Learning & AI
Web Development
Mobile Development
Backend & Systems
Trading & Finance
Key Achievements
500+ Downloads
Custom MT5 trading indicators trusted worldwide
99% Training Accuracy
Achieved in Deep RL for Atari Tetris using PPO
78.9% AUC
Medicare fraud detection model performance
Multi-Platform Development
Full-stack expertise across web, mobile, and AI systems
Services
Machine Learning & AI
Deep learning, reinforcement learning, and AI solutions for real-world problems including algorithmic trading and predictive analytics.
Web Development
Full-stack web applications using React, Flask, and modern frameworks with API integration and responsive design.
Mobile Development
Android applications using Java, Kotlin, and Jetpack Compose with modern Material Design principles.
Systems & Backend
Backend systems, microservices, and APIs using Java, Spring, C++, and modern development practices.
Algorithmic Trading
Custom trading indicators, financial algorithms, and trading system development with MT5 integration.
Data Science
Data preprocessing, analysis, and visualization with expertise in healthcare AI and fraud detection systems.
Portfolio
Deep RL for Atari Tetris
Actor-Critic (PPO) with CNN state processing. Achieved 99% training accuracy and optimized rewards.
Medicare Fraud Detection
Deep neural network for healthcare fraud (78.9% AUC). Preprocessing pipeline with 40+ features.
Handwritten Digit Recognition
MNIST classifier with 98% training / 96% test accuracy. Real-time digit classification.
Speech-to-Text & TTS Web App
Flask + Google Cloud APIs. Real-time STT/TTS, sentiment analysis, Docker, Cloud Run.
Movie Discovery (Veni Vici)
React + Streaming API. Smart filters, ban lists, and scrollable discovery history.
AI Agents Directory
React + Vite + Tailwind. Card-based resource board listing 10+ AI agents.
E-Commerce Course Platform
Android app with Java + Material Design. Online course marketplace with cart and filtering.
Weather Apps
Kotlin + Jetpack Compose. Real-time weather APIs with modern Material UI design.
Traffic Light Simulation
Educational Android app simulating semaphore systems with custom animations.
Bank Loan Optimization System
C++ + Pthreads simulation optimizing bank loan rates with statistical modeling.
Spring MVC CRUD App
Java + Spring MVC + PostgreSQL. REST API, Thymeleaf UI, Hibernate validation.
TVMaze Integration Service
Java Quarkus microservice + PostgreSQL. CRUD + Swagger/OpenAPI docs.
📊 AriPoint Trading Indicator
Custom MT5 trading indicator with integrated analytics. Trusted worldwide with 500+ downloads across 3+ indicators.
🤖 Advanced Tetris RL Demo
Watch an AI agent learn to play Tetris using cutting-edge reinforcement learning techniques
🎮 Live Training Session
📊 Real-Time Learning Metrics
Current Score
0
Lines Cleared
0
Reward Signal
0.00
Training Episode
1
Advanced RL Architecture
This implementation features a sophisticated Double Deep Q-Network (DDQN) with Prioritized Experience Replay (PER) and Genetic Algorithm optimization. The agent progresses through three distinct phases: Exploration (500 games), Exploitation (500 games), and Genetic Optimization. It learns from 6 key game state features: total height, bumpiness, holes, lines cleared, piece position, and pillar height.
Training Progress
Ready to start training...
🔬 Technical Implementation Details
🧠 Double DQN Network
Dual neural networks (Primary & Target) with 3 hidden layers (32→32→32 neurons), trained on 6-dimensional state space
🎯 Advanced State Representation
6-feature state vector: total height, bumpiness, holes, lines cleared, piece position, and pillar height analysis
⚡ Prioritized Experience Replay
Intelligent memory buffer with 30,000 experiences, prioritized by Temporal Difference error for optimal learning
🏆 Genetic Algorithm Optimization
Evolutionary strategy automatically optimizes reward function parameters through population-based selection and mutation
📈 Live Trading Analysis
Bitcoin H1 - Real Market Data
Loading real Bitcoin market data...
Trading Performance
Total Signals
0
Winning Trades
0
Losing Trades
0
Win Rate
0%
Total Profit
$0.00
Profit Factor
0.00
Technical Specifications
📊 Moving Average
SMA Period: 15 | Applied Price: Median
📈 Volatility Bands
ATR Period: 30 | Multiplier: 2.0
🎯 Risk Management
TP: 1.0× ATR | SL: 2.0× ATR
💰 Profit Levels
PP1: 1.6× ATR | PP2: 3.2× ATR
Custom MT5 Indicator Algorithm
This indicator uses real Bitcoin H1 market data from Binance API with dynamic volatility bands based on ATR (Average True Range) and a 15-period moving average. Signals are generated when price touches the bands with confirmation from candlestick patterns. The algorithm includes sophisticated risk management with multiple profit-taking levels.
Key Features
- Real Market Data: Live BTC/USDT from Binance API
- Dynamic Bands: ATR-based volatility measurement
- Signal Confirmation: Candlestick pattern validation
- Risk Management: TP/SL with partial profit levels
- Real-time Analysis: Live signal generation and tracking
Contact Me
temirlan9908@gmail.com
linkedin.com/in/temirlan-kdyrkhan
GitHub
github.com/TemirlanKN