CV
This is a description of the page. You can modify it in '_pages/cv.md'. You can also change or remove the top pdf download button.
Contact Information
| Name | Tunjay Gafarli |
| Professional Title | Compiler & GPU Engineer | CENG Undergrad |
| 4tvnexmgmt@gmail.com |
Professional Summary
Computer Engineering undergraduate specializing in compiler design, deterministic GPU execution, and memory-safe Edge AI. Creator of Vyne, a multi-target systems programming language.
Experience
-
2025 - Present Baku, Azerbaijan
Teaching Assistant (Computer Laboratory)
Türkiye-Azerbaijan University
Assisting students with low-level programming, computer architecture, and systems design.
- Managing lab sessions and helping students debug C/C++ and hardware projects.
- Grading assignments and providing technical feedback on systems architecture.
Education
-
2024 - 2028 Baku, Azerbaijan
B.S.
Türkiye-Azerbaijan University
Computer Engineering
- Compilers, Systems Architecture, Embedded Systems, Digital Signal Processing
- Teaching Assistant for the Computer Laboratory.
Projects
-
2025 - Present Vyne Compiler
A statically-analyzed systems programming language with a multi-target backend (x86, CUDA PTX, AVR, ESP32).
- Architected custom CUDA PTX kernels to achieve bit-exact deterministic GPU training, bypassing PyTorch’s atomicAdd non-determinism.
- Reduced peak VRAM for ESPCN neural network training by 94% (from 11.2 MB to 0.65 MB) using explicit @device region preallocation.
- Built a hardware abstraction layer (hardware.toml) that automatically resolves SPI display/touch presets for ESP32 deployment.
-
ESP32 Wi-Fi Radar & Audio Hardware
Bare-metal hardware projects built using the Vyne compiler backend.
- Developed a live RF packet monitor graphing Management/Data frames on a 3.5-inch SPI display in under 50 lines of Vyne code.
- Designing an integrated Smart Guitar Box using ESP32 for real-time DSP (FFT/Amp Profiling) and Arduino Uno R3 for hardware interrupts and UI.
Skills
Systems & Compilers (Advanced): C++, C, Compiler Design, LLVM, PTX, Memory Management, Region-based Allocation, Explicit Memory Models
GPU Architecture & CUDA (Advanced): CUDA Driver API, PTX JIT Compilation, Custom Kernel Design (Gather/Scatter), Warp-level Programming, VRAM Preallocation, Bypassing atomicAdd, Nsight (Basic)
Machine Learning Systems (Advanced): Source-to-Source Autodiff, Deterministic Training, PyTorch (Weight Parity/Export), Neural Network Inference (CNN, ESPCN), Custom Optimizers (AdamW, SGD), Edge AI, TinyML
Embedded & Hardware (Intermediate): ESP32, AVR (Arduino), SPI/I2C, Digital Signal Processing (DSP), Real-time Systems, RF Packet Monitoring
Languages
Azerbaijani : Native speaker
English : Fluent
Turkish : Fluent
Interests
Systems Engineering: Compilers, GPU Architecture, Deterministic ML, Edge AI
Music Production: Audio Engineering, Real-time DSP, Analog Hardware, 4+ Years Experience