I am a Computer Engineering graduate from Gazi University, where I graduated as both Faculty Valedictorian and Computer Engineering Department Valedictorian with a 3.93/4.00 GPA, ranking 1st out of 677 students across the Faculty of Engineering and 1st out of 120 students in my department.

I focus on building practical AI systems that connect models with real users, real data, and real operational constraints. My work spans production ML, MLOps, LLM applications, edge AI deployment, agentic RAG systems, and reliable backend services.

I am especially interested in designing AI products end to end: from model development and evaluation to deployment, monitoring, automation, and system reliability.

• AI Engineering: TensorFlow, PyTorch, scikit-learn, TensorFlow Lite, Hugging Face, LangChain, NLP, time-series models, federated learning
• MLOps & Backend: Docker, Flask, FastAPI, Socket.IO, Streamlit, REST APIs, data pipelines, CI-friendly testing, Git, Linux
• Languages & Data: Python, SQL, C++, pandas, NumPy, feature engineering, data validation, monitoring, dashboards
• Systems & Reliability: FMEA/FMECA, Fault Tree Analysis, safety-critical engineering workflows, automation tooling

• Built an automated AI-agent testing and preview module using Flask, GPT-4, internal APIs, and structured debugging.
• Reduced repeated agent validation time from approximately 10 minutes to 1 minute per knowledge, skill, or persona testing cycle.
• Refactored Model Context Protocol modules to improve extensibility, error handling, and retrieval reliability for new agent workflows.

• Developed Python/VBA automation tools for FMEA/FMECA consistency checks, reducing manual review effort by up to 99%+ in safety-critical workflows.
• Connected FMECA outputs with Fault Tree Analysis concepts to support failure propagation mapping and quantitative risk assessment.

• Preprocessed and modeled 40M+ rows of internal operational traffic data.
• Built data cleaning, validation, feature engineering, and time-series preparation workflows.
• Delivered forecasting results through web-based dashboards and interfaces.

A TÜBİTAK-supported high-dexterity prosthetic hand control system using deep learning for real-time sEMG gesture classification.

• Trained a DANN-based sEMG classifier for 29 hand gestures, reaching 96% accuracy.
• Converted the model to TensorFlow Lite for Raspberry Pi 5 deployment.
• Connected real-time predictions to a Flask/Socket.IO dashboard and PCA9685 servo control pipeline.
• Tech Stack: TensorFlow Lite, PyTorch, Raspberry Pi 5, Flask, Socket.IO, PCA9685, Python

A lightweight, Docker-first MLOps tool for detecting feature and concept drift in production ML pipelines.

• Implemented statistical drift tests, adaptive windowing, alert hooks, diagnostic plots, and synthetic stream generation.
• Added model registry updates and retraining trigger logic for production monitoring workflows.
• Designed the project as a modular CLI-first tool for practical ML observability.
• Tech Stack: Python, Docker, NumPy, SciPy, River, Loguru

A Dockerized prototype for event-driven agentic RAG workflows and drift-aware MLOps design.

• Built an event producer, processing agent, incremental vector-store update flow, local embeddings, and metrics layer.
• Exposed /search and /ask APIs for retrieval and question-answering workflows.
• Designed the system around modular ingestion, retrieval, monitoring, and future cloud deployment patterns.
• Tech Stack: Docker, LangChain, FastAPI/Flask, vector stores, local embeddings, AWS blueprint design

A decentralized federated learning prototype with blockchain-backed model update auditability.

• Built a scalable federated learning prototype with Flask/Socket.IO visualization.
• Tested on a 3-client, 10-class non-IID setup where the aggregated central model reached approximately 87% accuracy, compared to 28–38% local client accuracy.
• Used blockchain-backed update logging to improve transparency and auditability in decentralized ML workflows.
• Tech Stack: Python, PyTorch, Flask, Socket.IO, Solidity/Web3 concepts

I actively contribute to AI and ML infrastructure projects, with focused pull requests under review for:

• Apple MLX
• Google TensorFlow
• PyTorch Lightning
• TruLens

My contributions focus on scoped reliability improvements, regression tests, local validation commands, edge-case behavior, benchmark diagnostics, and clear PR documentation.

• Email: nizamfurkanegecan@gmail.com
• LinkedIn: linkedin.com/in/furkan-egecan-nizam
• GitHub: Explore more of my AI, MLOps, and infrastructure projects here on my profile.