Metehan KIZILCIK
About
I am a Software Engineering student with a strong foundation in object-oriented programming, clean code, software architecture, design patterns, and data structures. I build AI-powered full-stack applications using AI agents, automation tools, modern web technologies, and generative AI integrations.
My long-term focus is AI Engineering, MLOps, Machine Learning, Deep Learning, RAG systems, AI harnesses and agentic workflows, scalable inference architectures, and the engineering practices required to take intelligent systems from experimentation to reliable production at scale.
Professional Experience
Technical Lead — AI & Fullstack Engineer, Raskan Ajans
Lead technology strategy and full-stack development as CTO, overseeing AI-driven product development for the agency. Delivered and contributed to 50+ real-world projects for a diverse range of clients across different industries, business types, and digital needs.
AI Engineer & Project Lead, BLUESENSE (Smart Beauty)
- Worked across Computer Vision, agentic chatbot systems, embedding-based product recommendation, Vision Transformers, classical deep-learning architectures, data scraping and labeling, model training, Dockerization, and live inference deployment.
- Led the Computer Vision workstream for a production acne-analysis feature and coordinated a team of roughly 20 people from data collection through production.
- Fine-tuned DINOv3 ViT-H+ with LoRA for six acne subtypes, reaching a 97% Top-2 F1 after 11 iterative model versions.
- Designed a production inference pipeline covering detection, acne/non-acne classification, subtype classification, severity scoring, and ensemble logic.
- Built the SmartChat agent-based chatbot architecture, including live web-search integration, and designed a semantic product-matching engine using vectorized product data.
Digital Content Creator, Instagram (@mo_tunn1)
Built a tech/project-focused personal channel to 7,000+ followers and 1.5M+ views within 1.5 months.
Selected Projects
End-to-end YouTube channel intelligence platform that turns channel, video, and comment data into sentiment analysis, recurring themes, audience profiles, and brand/product partnership recommendations. Includes asynchronous jobs, progress tracking, caching, and multi-stage AI fallback logic.
Cross-platform desktop AI agent for Windows, macOS, and Linux with Claude, OpenAI, Gemini, Groq, OpenRouter, and Ollama support. Includes step-by-step planning, screenshot/OCR context, pointer guidance, browser automation plugins, voice I/O, secure credential storage, and automated CI/CD.
Open-source package and local MCP server for packing long documents, codebases, and PDFs into token-budgeted, evidence-dense context for LLMs. A benchmark study reported up to 74.6% context-token reduction, 3.90× latency speedup, and a 15.6% relative accuracy lift over full-context prompting.
IoT voice-translation device using a distance sensor, I2S microphone, I2S amplifier, and a three-stage cloud AI pipeline for speech-to-text, translation, and text-to-speech. Includes a browser-based control panel served directly from the device.
Service-Oriented student mentoring platform with mentor/student panels, scheduling, task assignment, exam-result tracking, JWT authentication, a mock SOAP identity service, and a Scikit-learn Gradient Boosting model served through gRPC.
Full-stack spaced-repetition vocabulary app using a six-interval reinforcement model, personalized AI learning suggestions, generated stories/images, a Wordle-style review game, and SonarQube-based code-quality checks.
End-to-end machine-learning pipeline for classifying computer-science academic abstracts as human-written or AI-generated. Benchmarked eight classical ML algorithms and reached approximately 96% accuracy with Random Forest.
Skills
Languages & Engineering: Python, SQL, NoSQL, Node.js/Express, React, C++, embedded systems, full-stack web development, OOP, clean code, design patterns, data structures and algorithms.
AI / ML: Machine Learning, Deep Learning, Computer Vision, LoRA fine-tuning, Scikit-learn, TF-IDF, Sentence Transformers, embeddings, semantic search, RAG, AI agents, generative AI integration, chatbot architecture, multi-provider LLM orchestration, MLOps fundamentals.
Infrastructure & Data: Cloud, Docker, Git, GitHub Actions CI/CD, MongoDB, PostgreSQL, MySQL, relational and vector databases.
APIs & Architecture: gRPC, REST, SOAP, JWT authentication, Service-Oriented Architecture, MCP (Model Context Protocol).
Education & Certifications
Bachelor's Degree, Computer Software Engineering
Manisa Celal Bayar University · GPA: 3.77
- Cybercrime and Security Specialist — Istanbul Esenyurt University
- Machine Learning Specialization — DeepLearning.AI / Stanford Online
- Advanced Learning Algorithms — DeepLearning.AI / Stanford Online
- Supervised Machine Learning: Regression and Classification — DeepLearning.AI / Stanford Online
- AI Agents Fundamentals — Hugging Face