name: Mehmet Gümüşlocation: Istanbul, Turkey 🇹🇷current_roles:
- AI Trainer @ Invisible Technologies
- Embedded SW Team Lead @ UKET (Volunteer)philosophy:
- "Physics first, code second"
- "Simulate before you build"
- "Measure everything, assume nothing"focus_areas:
- Aerospace & Defense Software
- AI/ML & LLM Training (RLHF)
- Real-time Physics Simulations
- Embedded Systems & Robotics
- 3D Visualization & WebGLfun_fact: "I build rockets in software before they fly in reality 🚀"
| Role | Organization | Period | Highlights |
|---|---|---|---|
| 🤖 AI Trainer | Invisible Technologies | 2025 - Now | LLM validation, RLHF, Turkish NLP |
| 🚀 Founder & Developer | Bilir.app | 2024 - Now | Flutter, OOP design, 6-person team lead |
| ⚙️ Embedded SW Lead(Volunteer) | UKET | 2023 - Now | Rover systems, sensor fusion, C/C++ |
| 🔬 Software Intern | TÜBİTAK MAM | 2023 | DSP, VNA data, MATLAB/Python |
Rocket Engine & Flight Simulation Platform
🔬 NASA CEA methodology with Gordon-McBride equilibrium solver
⚗️ Gibbs free energy minimization for combustion analysis
🚀 6-DOF flight simulation with quaternion-based orientation
📊 Monte Carlo dispersion analysis for landing predictions
❄️ Regenerative cooling thermal analysis (Bartz correlation)
🎯 Multi-stage vehicle support (Falcon 9, Saturn V presets)
⚡ Validated against NASA CEA with <2% error • Numba JIT for 10-100x speedup
🧪 226 automated tests with pytest • Published on PyPI
Advanced Real-Time Satellite Tracking & Orbital Analysis Platform
🌍 25,000+ satellites & debris real-time tracking with SGP4 propagation
📡 Doppler Shift calculation for radio frequencies
📉 Orbital Decay prediction algorithms⚠️ Conjunction analysis for collision warnings
📍 Pass Prediction algorithm for observer locations
📱 PWA support & AR mode with compass-based guidance
High-Fidelity NEO (Near-Earth Object) Defense Simulator
🔬 N-Body gravity simulation with Velocity Verlet integrator (Rust)
🛰️ NASA NeoWs API integration for real asteroid data
📊 Monte Carlo collision probability analysis
💥 Kinetic Impactor deflection scenario modeling
🎬 Cinematic physics-based 3D rendering with Three.js
IEEE Std 686-2008 Compliant Radar Simulation & Operator Console
📡 Swerling I-IV fluctuation models & ITU-R P.676 atmospheric attenuation
🎯 LFM/Barker waveforms, CA-CFAR thresholding, SAR/ISAR imaging
🤖 Random Forest ML for target classification (Drone/Fighter/Missile)
🖥️ 30+ FPS PPI Scope, A-Scope & Range-Doppler displays
🌐 Cross-platform builds (Windows, macOS, Linux) with GitHub Actions CI/CD
Real-time FDTD Electromagnetic Solver on WebAssembly
🌊 Yee Lattice Algorithm running directly in-browser with zero-copy memory access
🛡️ CPML Boundaries (Convolutional Perfectly Matched Layer) for open-space simulation
🔬 Interactive Lab with 7 scenarios (Double Slit, Waveguide, Lens, etc.)
📈 Real-time Signal Analysis with virtual oscilloscope & energy divergence monitoring
⚡ Performance: 60+ FPS on 512x512 grid using custom Rust physics engine
Full-Stack Turkish Lunar Mission Simulator
20+ REST API endpoints • Real-time 3D orbital mechanics • Integration testing
WebGL Missile Guidance & 6-DOF Simulation
RK4 physics integration • Proportional Navigation guidance • Real-time control loop
Interactive Geography Game Platform
Google Maps/Leaflet integration • XP/Level algorithms • Local caching optimization
| Principle | Description | |
|---|---|---|
| 🎯 | Design First | Understand the problem before writing code |
| 🔬 | Simulate | Physics-based models over black-box assumptions |
| 📊 | Measure | Quantified uncertainty & error bounds |
| ⚡ | Profile | Benchmark critical paths |
| 🛡️ | Fail Safe | Graceful failure, never silent |
| Domain | Technologies |
|---|---|
| Physics | 6-DOF dynamics, RK4/Verlet integration |
| State Est. | Kalman Filter, EKF, UKF |
| Control | PID, Adaptive, Gain Scheduling |
| Analysis | Monte Carlo, Genetic Algorithms, Pareto |
| Guidance | Proportional Navigation, Missile dynamics |
| DSP | Monopulse tracking, Doppler, ECM |
Core Competencies:
- 🔹 6-DOF rigid body dynamics & kinematics
- 🔹 Sensor fusion & calibration (SOLT method)
- 🔹 Real-time control loop design
- 🔹 Hardware-in-the-loop (HIL) testing mindset
- 🔹 Monte Carlo uncertainty propagation
| Domain | Focus |
|---|---|
| LLM | RLHF, Evaluation pipelines, Hallucination analysis |
| ML | KNN, SVM, Decision Trees, Classification |
| Optimization | Genetic Algorithms, Multi-objective (Pareto) |
| Data | Feature engineering, Synthetic data generation |
| Area | Technologies |
|---|---|
| MCU | ESP32, Arduino, ARM Cortex-M, Raspberry Pi |
| Protocols | UART, I2C, SPI, CAN |
| Control | Adaptive PID, Gain Scheduling, Flight Controllers |
| Tools | Altium Designer, Logic Analyzer, RTOS concepts |
Highlights:
- 🔹 Rover & robotic arm embedded architecture
- 🔹 Motor driver control optimization (C/C++)
- 🔹 PocketVNA signal processing
- 🔹 FPV drone flight controller logic
🧠 AI & Data Science
🌐 Frontend & 3D
💻 Backend & Desktop
⚙️ Embedded & Low-Level
🔧 DevOps & Tools
| Certificate | Institution |
|---|---|
| 🛰️ Digitalisation in Aeronautics & Space | TUM (Technical University of Munich) |
| 📡 IoT Networking(Honors) | University of Illinois |
| 💻 IT Essentials: PC Hardware & Software | Cisco Networking Academy |
| Skill | Level |
|---|---|
| 🧠 AI/ML & LLM | ⬛⬛⬛⬛⬛⬛⬛⬛⬛⬜ 95% |
| 🚀 Simulation & Physics | ⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛ 97% |
| 💻 Full-Stack Dev | ⬛⬛⬛⬛⬛⬛⬛⬛⬛⬜ 90% |
| ⚙️ Embedded Systems | ⬛⬛⬛⬛⬛⬛⬛⬛⬜⬜ 85% |
| 📡 Radar & DSP | ⬛⬛⬛⬛⬛⬛⬛⬛⬜⬜ 80% |
| 🎨 3D/WebGL | ⬛⬛⬛⬛⬛⬛⬛⬛⬛⬜ 90% |
| 📱 Mobile (Flutter) | ⬛⬛⬛⬛⬛⬛⬛⬛⬜⬜ 85% |
| Language | Level |
|---|---|
| 🇹🇷 Türkçe | Native |
| 🇬🇧 English | Intermediate (B1-B2) |
"First, solve the problem. Then, write the code." - John Johnson
⭐ Built with 🚀 by SpaceEngineerSS