// PROJECT PORTFOLIO

ELIFNUR TABALA

Full technical deep-dive: every project, every stack decision, every GitHub link. No fluff.

ROS2 ✓ KTR 87/100 ★ TEKNOFEST SPRING BOOT C / SYSTEMS CLAUDE API GPA 3.32
▶ PROJECTS LOADED: 5  ·  COMMITS: MANY  ·  BUGS FIXED: MORE
87 KTR SCORE / 100
3.32 GPA / 4.00
679 TESTS PASSING
10+ GAZEBO SCENARIOS
4 LANGUAGES CODED
// COMPLETED LEVELS

PROJECT_ARCHIVE

★ TEKNOFEST KTR: 87/100
🚗
LEVEL 05 — FLAGSHIP PROJECT
SAHI OTONOM
ROBOTAKSI
Software Developer · Autonomous Systems · 2025–Present

Autonomous vehicle system competing at Teknofest Robotaksi — Turkey's premier autonomous driving competition. Full perception, navigation, and control stack on real hardware. KTR (technical report) scored 87/100.

PERCEPTION LiDAR + ZED 2i stereo camera sensor fusion pipeline in ROS2 Humble; real-time depth estimation and obstacle detection
LANE DETECT YOLOPv2-based lane detection integrated with OpenCV; output fed as ROS2 topic to decision node
NAVIGATION A* path planning algorithm; PID control loop for steering, braking, propulsion
SIMULATION 10+ autonomous driving test scenarios in Gazebo before hardware deployment; caught critical pipeline errors pre-integration
HARDWARE STM32F4 MCUs over I2C (400kHz) for steering, braking, propulsion; Ubuntu 22.04, ROS2 Humble
TEAM 4-engineer robotics software team; collaborative ROS2 node development and integration
ROS2 HUMBLEPYTHONOPENCV YOLOPV2LIDARZED SDK GAZEBOA*PID STM32F4I2CUBUNTU 22.04
⌥ GITHUB (PRIVATE) ↗ TEKNOFEST
👗
LEVEL 04 — AI VISION
DRAPE
AI OUTFIT STUDIO
Full-Stack Developer · AI/CV · Solo Project

Real-time virtual try-on system using computer vision and AI. Upload a garment — Claude Vision analyzes it, MediaPipe overlays it on your live camera feed, emotion detection tells you if you actually like it.

CV PIPELINE MediaPipe pose estimation + body segmentation to overlay garments on live video feed in real time
AI LAYER Claude Vision API for intelligent garment image processing and analysis; remove.bg for background removal
EMOTION face-api.js facial emotion detection to score user satisfaction with outfit; drives recommendation engine
BACKEND Node.js orchestration layer managing AI/ML pipelines across Claude Vision, MediaPipe, and remove.bg APIs
NODE.JSCLAUDE VISION API MEDIAPIPEFACE-API.JS JAVASCRIPTREMOVE.BG APIAR
⌥ GITHUB
🛒
LEVEL 03 — BACKEND
HASAD
DIGITAL MARKETPLACE
Backend Developer · Java · Spring Boot

Production-grade multi-vendor marketplace backend for local producers. Full auth, role management, cart, orders, reviews — 679 tests passing, deployed with Docker.

AUTH JWT-based authentication; role-based authorization across 4 roles (SUPER_ADMIN, ADMIN, PRODUCER, CONSUMER)
DATABASE PostgreSQL with normalized multi-tenant schema; Flyway migrations (V1–V5); multi-tenant data isolation
API RESTful API with Spring Boot 3.2; full admin panel (8+ pages), cart with live badge, address CRUD, review system
TESTING 679-line test suite passing; fixed critical LOWER() bytea bug via Java stream filtering
INFRA Dockerized deployment; Spring Security; port 8080 /api
JAVA 17SPRING BOOT 3.2 POSTGRESQLJWT FLYWAYDOCKER SPRING SECURITYREST API
⌥ GITHUB
🖥️
LEVEL 02 — SYSTEMS PROGRAMMING
PAGE REPLACEMENT
SIMULATOR
Systems Programmer · C · OS Internals

Low-level memory management simulation in C. Implemented FIFO, LRU, and Optimal from scratch, benchmarked across configurable frame sizes, measured real page fault deltas.

ALGORITHMS FIFO, LRU, and Optimal page replacement — implemented from scratch in C with no library dependencies
BENCHMARKS Compared page fault rates across 3 configurable frame sizes; Optimal ~30% fewer faults vs FIFO (verify against output)
ENGINE Simulation engine with configurable frame count for OS memory management analysis across access patterns
WHY IT MATTERS Demonstrates real OS internals knowledge — not a framework tutorial. Same memory management principles used in embedded/RTOS systems
CSYSTEMS PROGRAMMING OS INTERNALSMEMORY MANAGEMENT
⌥ GITHUB
🍽️
LEVEL 01 — SHIPPED TO PRODUCTION
ŞANLIURFA KEBAP
RESTAURANT SITE
Full-Stack Developer · Frontend · Deployed

Designed and deployed a production website for a real local restaurant. Responsive UI, menu management, live in production. Real users, real traffic.

STACK HTML, CSS, JavaScript — zero framework dependencies, hand-crafted responsive layout
STATUS Live at sanliurfakebap.com — real production deployment, real users
HTMLCSS JAVASCRIPTDEPLOYED ✓
↗ LIVE SITE ⌥ GITHUB
// SKILL TREE

TECH_STACK

LANGUAGES
Python
Java
C
JavaScript
C++
ROBOTICS & VISION
ROS2 Humble
OpenCV
LiDAR / ZED SDK
Gazebo
Sensor Fusion
BACKEND
Spring Boot 3.2
PostgreSQL
Node.js
Docker
REST API / JWT
EMBEDDED & CONTROL
I2C / UART / SPI
STM32F4
PID Control
A* Path Planning
Linux (Ubuntu)
// INSERT COIN TO CONTINUE

LET'S BUILD

Open for internships, collabs, and anything sufficiently interesting.
Response time: faster than a ROS2 topic callback.

✉ EMAIL ⌥ GITHUB ↗ LINKEDIN ← MAIN SITE