Skip to content
View overdoseparsa's full-sized avatar

Block or report overdoseparsa

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.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. 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
overdoseparsa/README.md

Hi 👋, I'm Parsa Khaki

Backend Software Engineer · Python · FastAPI · Django · Computer Vision


  • 🔭  Currently building auth-service-fastapi — a production-grade auth microservice with JWT token family tracking, Redis blacklisting, and async SQLAlchemy 2.0
  • 👁  Building real-time computer vision systems — multi-camera RTSP license plate recognition (YOLO + OCR) with a shared-memory frame pipeline and a central GPU inference scheduler
  • 🛠  I care about connection pool hygiene, atomic transactions, and layered architecture — not just making things work
  • 🌱  Going deeper into distributed systems, async patterns, microservice design, and production AI/vision pipelines
  • 💬  Ask me about FastAPI, Django, PostgreSQL, JWT security, YOLO / ANPR / RTSP pipelines
  • 📫  Reach me at parsakhakiy@gmail.com


🧱 What I focus on

interests = [
    "Production-grade backend systems",
    "Async Python (asyncpg, SQLAlchemy 2.0, aioredis)",
    "Auth & security design (JWT, token rotation, reuse detection)",
    "Clean architecture & layered design",
    "PostgreSQL performance & transaction management",
    "Computer vision in production (YOLO, ANPR/LPR, RTSP streaming, GPU inference)",
]

👁 Computer Vision & AI

  • Multi-camera LPR pipeline — redesigned RTSP ingestion for a dynamic number of cameras (driven by JSON config, nothing hardcoded):
    • capture processes/threads per camera group, with a consumer per capture process
    • per-camera multi-slot ring buffer in shared memory (frames / sequence / metadata segments, seqlock-based) to avoid copying frames between processes
    • a single central YOLO process on GPU that loads the model once and serves all cameras
    • benchmarked different process × thread layouts (1×30, 2×15, 3×10, 5×6, 6×5) to find the best CPU/GPU balance
    • detection, OCR, event processing and output integrations kept intact through the redesign
  • VizGuard — vehicle plate detection and blacklist management app.
  • Exploring
    • A brand-agnostic industrial event platform built on IVS camera analytics (PPE, intrusion, forklift–human proximity, ANPR…), aggregating events across sites and camera brands into one dashboard
    • AI traffic incident detection using existing road cameras to detect accidents and alert emergency services
    • RAG chatbots as a SaaS for small Persian-speaking businesses

🚀 Featured Projects

Project Stack Description
auth-service-fastapi FastAPI · PostgreSQL · Redis · asyncpg JWT auth microservice with refresh token rotation, family-based reuse detection, and Redis blacklisting
LPR Pipeline (private) Python · YOLO · OpenCV · RTSP · CUDA Multi-camera license plate recognition with shared-memory ring buffers and central GPU scheduling
VizGuard (private) Python · Computer Vision Vehicle plate detection & blacklist management
Neutrino Python · Docker DevOps-oriented Python project
Django-Ci-CD Django · GitHub Actions CI/CD pipeline configuration for Django

🛠 Languages and Tools

python  fastapi  django  postgresql  redis  docker  git  linux


📬 Connect

📧 parsakhakiy@gmail.com

Pinned Loading

  1. auth-service-fastapi auth-service-fastapi Public

    FastAPI Authentication Service - For Production Ready

    Python 1

  2. Django-assignment-danaJoyan Django-assignment-danaJoyan Public

    this is assignment for dana joyan

    Python

  3. Neutrino Neutrino Public

    This project is for DevopsNeutrino

    Python 2

  4. Django-Ci-CD Django-Ci-CD Public

    The django CI CD configuration

    JavaScript 1