Skip to content

I am an AI Engineer. I build multi-agent systems. I ship RAG pipelines

AI Engineer building reliable LLM applications, multi-agent systems and RAG pipelines.

Contact →

a page from my notebook

About

hi! thanks for stopping by my notebook.

I'm Beyza Nur, a Computer Engineer from MEF University and an AI/ML Engineer at Mindsane, an Istanbul-based AI consultancy. For the past year I've been building LLM-based systems end to end: multi-agent assistants with LangGraph, RAG pipelines over messy real-world documents, and ML models behind two papers accepted at IEEE ASYU 2026. I care about the parts that make AI actually work outside a notebook: evaluation, guardrails and deployment on AWS.

Right now I'm building SkyTrace, a drone flight-log analysis agent, and a multi-agent IT support assistant. Next, I want to join a team shipping AI products at scale, where reliability matters as much as the demo.

Education: B.Sc. Computer Engineering, MEF University (100% English)

things I've built

Projects

View all repositories on GitHub →
  • SkyTrace analysis view: a timeline of flight modes and findings, a list of motor and vibration findings, and the motor-output plot behind one saturation warning.
    In ProgressOct 2026 – Present

    SkyTrace

    Drone flight-log analysis — parses PX4/ArduPilot logs, detects anomalies and answers questions with a LangGraph agent whose every claim is checked against tool outputs.

    • Python
    • pandas
    • LangGraph
    • LiteLLM
    • Ollama
    • Qdrant
    • bge-m3
    • FastAPI
    DetailsCodeDemo
  • Assistant chat in Turkish: the supervisor splits a request into a VPN and an e-mail issue, the Triage and Solution agents call their tools, and the answer lists fix steps with the knowledge-base source.
    In Progress2026 – Present

    Multi-Agent IT Support Assistant

    A helpdesk assistant for Turkish IT requests — a LangGraph supervisor hands off to Triage, Solution, Ticket and Analytics agents; tickets open only after user approval.

    • LangGraph
    • FastMCP
    • pgvector
    • sqlglot
    • Presidio
    • BERTurk
    • FastAPI
    • k3s
    DetailsCode
  • Doggo assistant (Turkish UI): asked for a walker in Beşiktaş, it lists 10 walkers on an Istanbul map with their rating, years of experience and hourly price.
    CompletedCapstone ProjectTeam projectSep 2025 – Jun 2026
    code under NDA

    Doggo — Multi-Agent Chatbot

    Capstone project — turned the Doggo pet walker app into an AI-powered assistant with three agents coordinated by an orchestrator. Team of three.

    • multi-agent
    • orchestration
    • tool usage
    • multi-step reasoning
    DetailsDemo
  • EduLift chat: a teacher asks for a quiz for Machine Learning 101 and the assistant, built on Microsoft Copilot, asks for the lesson name to tailor it.

    Hackathon 1st Place

    CompletedTeam projectFeb 2025

    EduLift

    AI-powered education assistant built in 2 days — automated grading, lesson planning and real-time learner analytics. 1st place at Miami Horizon AI Hackathon.

    My part: Involved across the whole project — lesson planning, quiz generation and automated grading, and learner analytics.

    • Next.js
    • Tailwind
    • Microsoft Copilot
    • Azure
    DetailsCodeDevpost

More projects

  • Team projectInternship CapstoneCompleted

    DocuChat — Multimodal PDF Q&A

    RAG assistant over complex PDFs (text, tables, figures) that highlights the cited regions on the page. Team project — I led PDF processing and integration.

    my part: Led the PDF processing module and system integration.

    • Qwen-VL
    • Pydantic
    • Pinecone
    • MiniLM
    DetailsCode
  • Completed

    Brain MRI Classification & Clustering

    End-to-end ML pipeline on 6,056 brain MRI images with a 55-feature vector; XGBoost reached 91.67% accuracy (AUC 0.985).

    • Python
    • scikit-learn
    • XGBoost
    • GLCM
    DetailsCode
  • Completed

    Stock Management System (Flo)

    Full-stack stock management app built during my Flo Group internship — React front end, Spring Boot REST API with JWT login, and PostgreSQL.

    • React
    • Spring Boot
    • Java
    • PostgreSQL
    DetailsCode
  • Completed

    Grade Release System

    Cloud Engineering course project — a REST API on AWS in three stages (EC2 → ALB + Auto Scaling → Lambda + API Gateway), load-tested with k6 up to 1,650 req/min.

    • AWS
    • EC2
    • ALB
    • Auto Scaling
    Details
  • Team projectApp JamCompleted

    DoğaDostum

    AI-powered mobile app for plant and mushroom identification, built in a 3-day App Jam at the Google Game and Application Academy.

    my part: Built the image-based plant and mushroom classification feature and integrated it into the app.

    • image classification
    • mobile app
    DetailsCode
View all repositories on GitHub →

where I've worked

Experience

  1. AI/ML Engineer @ Mindsane

    Oct 2025 – Present

    • Developed machine learning pipelines for churn prediction in a subscription streaming service, including feature engineering, model evaluation, and performance analysis.
    • Built a multi-agent application using Python and LangGraph, implementing task orchestration and tool integration.
    • Contributed to research on sequential learning and customer retention, resulting in a paper accepted at IEEE ASYU 2026.

  2. Software Engineer Intern @ Flo Group

    Aug 2025

    • Developed a full-stack stock management application using React, Spring Boot, and PostgreSQL.
    • Worked on application functionality and collaborated with other departments to refine requirements and improve usability.
    → Stock Management System (Flo)→ code

  3. AI/ML Engineer Intern @ Eve Mağazacılık

    Jul 2025

    • Developed AI-driven chatbots and data-driven systems; explored Reinforcement Learning approaches for churn reduction, contributing to real-world production applications.

  4. AI/ML Engineer Intern @ Mindsane

    Jul 2025

    • Built and optimized RAG-based chatbots and ML prototypes in Python, leveraging vector databases and PDF pipelines to deliver practical solutions.
    → DocuChat — Multimodal PDF Q&A

Leadership & Volunteering

  • Vice President @ Ivy in MEF

    Leadership

    Oct 2022 · 11 mos

    Led a 35-member team, managing task distribution, coordinating weekly meetings, and liaising with the Dean's office.

  • Volunteer @ Ders Ortağım

    Volunteer

    Nov 2023 · 4 mos

    Provided in-class and out-of-class academic support to visually impaired learners.

hover or tap a logo to see where I used it

AI & Tech Stack

drag to spin ↻

Core:
LangGraph, RAG, PyTorch, FastAPI, AWS, Docker
Also used:
Python, LangChain, MCP, Hugging Face, Ollama, Qdrant, scikit-learn, pandas, NumPy, OpenCV, CUDA, ONNX Runtime, Gradio, PostgreSQL, pytest, DynamoDB, k3s, Prometheus, Grafana, MLflow, GitHub Actions, Git, GitHub, Linux, k6, Wireshark, React, Next.js, TypeScript, Flutter, Spring Boot, Java, C, Oracle
See full stack

ML / DL

Frontend & Mobile

say hi!

Contact

Open to AI Engineer roles. The fastest way to reach me is e-mail or LinkedIn.