Skip to content
View Logshi's full-sized avatar
🎯
Focusing
🎯
Focusing

Block or report Logshi

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Logshi/README.md

Şirvan Kesici

Edge AI Engineer Β· Computer Vision Β· Embedded Systems Β· C++ / Python / Rust / Flutter

LinkedIn Email X YouTube


πŸ‘¨β€πŸ’» About Me

I'm an engineer with an Electrical & Electronics Engineering and Economics background, working at the intersection of Edge AI, computer vision and embedded systems. I like building the whole pipeline β€” from the sensor and the model to the interface that makes it useful.

  • πŸ”­ Currently working at SophtRun, building computer vision and AI-integrated systems
  • πŸ€– Focused on Edge AI: real-time inference on Raspberry Pi 5 + Hailo AI HAT+ (CPU vs NPU)
  • πŸ“· Building RTSP camera analytics, zone-based detection and dwell-time analysis in C++ / OpenCV
  • πŸ“± Bringing models on-device with Flutter + TensorFlow Lite
  • πŸ“Š Bridging economics and software through open-data projects (TCMB EVDS)
  • πŸ“« Reach me at sirvan@sophtrun.com.tr

πŸš€ Featured Projects

🎯 zone-analysis

RTSP/video dwell-time person detection with C++ + YOLO ONNX + OpenCV

  • Detects people inside user-defined polygon zones
  • Flags dwell-time threshold breaches, saves alert snapshots
  • CSV event logging for downstream analysis

🧠 hailoai

Real-time RTSP AI inference benchmark on Raspberry Pi 5 + Hailo AI HAT+

  • CPU vs NPU comparison on edge hardware
  • Latency, FPS and thermal behavior observation
  • Edge inference pipeline tuning

🎈 skyvision-control-case-study

Sanitized case study of a balloon flight safety & telemetry platform

  • ESP32 sensor module (GPS, IMU, barometer, temperature)
  • Flutter Android tablet app + RTSP camera monitoring
  • On-device AI vision modules and a live telemetry dashboard

πŸ“Š MacroTR

Open-source Turkish macroeconomic dashboard with FastAPI + React + PostgreSQL

  • TCMB EVDS macroeconomic data ingestion
  • Full-stack: API, relational storage, dashboard frontend
  • Where my economics background meets software engineering

On-device fire/smoke detection with Flutter + TFLite + YOLO11n

  • Real-time camera and single-image detection
  • Model conversion from PyTorch (.pt) to TFLite
  • Fully on-device inference (no cloud)

πŸ—£οΈ GM VoiceAI

Local Turkish voice AI assistant built with Python

  • Speech-to-text β†’ command handling β†’ text-to-speech
  • Local / privacy-friendly approach

πŸ› οΈ Tech Stack

Languages

C++ Python Dart JavaScript MATLAB

Computer Vision & AI

OpenCV ONNX TensorFlow YOLO

Edge & Embedded

Raspberry Pi Hailo ESP32 Linux

Backend & Mobile

FastAPI PostgreSQL Flutter React

Tools

Git GitHub CMake


πŸ“Š GitHub Stats

Stars Followers


Open to Edge AI, computer vision and embedded software opportunities.

⭐ If you find my projects useful, feel free to star them!

Pinned Loading

  1. hailoai hailoai Public

    Raspberry Pi 5 + Hailo AI HAT+ benchmark for real-time RTSP AI inference: latency, FPS and thermal testing.

    Python 7

  2. GM_Voiceai GM_Voiceai Public

    Local Turkish voice AI assistant built with Python.

    Python 3 1

  3. MacroTR MacroTR Public

    Open-source Turkish macroeconomic dashboard using TCMB EVDS data, FastAPI, React and PostgreSQL.

    Python 2

  4. Flutter-YOLO11n-Fire-Smoke-Detection-TFLite-Android- Flutter-YOLO11n-Fire-Smoke-Detection-TFLite-Android- Public

    Flutter Android app for on-device YOLO11n fire and smoke detection with TensorFlow Lite.

    C++ 1

  5. yolo_flutter_object_detection yolo_flutter_object_detection Public

    A real-time object detection app built with Flutter and powered by YOLO (You Only Look Once) models using TensorFlow Lite. The project demonstrates live camera inference on Android tablets/phones w…

    1

  6. skyvision-control-case-study skyvision-control-case-study Public

    Sanitized case study for a hot-air balloon telemetry, camera monitoring and safety platform.