Meftun Akarsu
I work on AI systems — LLM infrastructure, computer vision, and data pipelines. Three years at Bosch building inspection systems, anomaly detection, and process automation for manufacturing. Studying AI Engineering at TH Ingolstadt, Germany.
M.Sc. AI Engineering for Autonomous Systems, Technische Hochschule Ingolstadt (THI)
B.Sc. Electrical and Electronics Engineering, Uludag University (3.48 / 4.0)
Erasmus+ Exchange Program, University of Zilina (3.88 / 4.0)
Experience
Built Odigos, a production RAG chatbot for autism spectrum families in Italy (Fondazione Paideia). End-to-end pipeline: knowledge graph traversal + vector search + semantic reranking with Cohere and GPT-4.1. Deployed on Azure Container Apps with Neo4j, pgvector, CI/CD via GitHub Actions.
Fine-tuned Stable Diffusion (WAN 2.1) for Img2Vid pipelines. Built Graph-RAG and LightRAG systems with Neo4j. Managed Azure deployments and built n8n-based agent workflows for unstructured data extraction.
Built YOLO + PyTorch inspection pipelines that cut destructive testing from 9/day to 3/day. Automated welding parameter optimization with XGBoost. 5000 hrs/yr productivity gain via RPA.
Implemented RTLS (UWB, Bluetooth) for asset tracking. Built operator tracking with Raspberry Pi 4, YOLO, TensorFlow, and MediaPipe. Predictive maintenance models with XGBoost and LightGBM.
Built an OCR-based product recognition system integrating AI/ML with PLC (Snap7) for production line automation.
Selected Work
Odigos — RAG Chatbot
Production RAG chatbot for autism spectrum families in Italy. 58+ features, knowledge graph + vector search + semantic reranking on Azure with Neo4j and pgvector.
Visual Quality Control
YOLO + PyTorch inspection pipeline at Bosch. Cut destructive testing from 9/day to 3/day. OCR and Cognex vision inspections reduced customer complaints by 10%.
Cinematic Video Generation
link →Fine-tuned Wan2.1 I2V-14B with LoRA for cinematic scene synthesis from limited data. Single-GPU pipeline producing coherent 720p video sequences.
Aerial Object Detection
link →Compared YOLO v3-v8 on VisDrone-2019 aerial images. Optimized with TensorRT for Jetson Nano and Jetson Orin. Published at IEEE INISTA 2023.
Writing
RAG from First Principles — a 20-part build-it-from-scratch course Start at Part 1 →Essays
-
When the Co-Pilot Is an Algorithm: How Europe Plans to Make Aviation AI Trustworthy
New — most recent essay10 min read EN · TRThe EU's aviation regulator just published a 239-page concept paper on making AI safe enough to fly. From artificial narrow intelligence to learning assurance and the W-shape process, the mental model behind trustworthy aviation AI, explained for non-experts.
- #EASA
- #Aviation
- #AI Safety
-
ASELSAN TOYGUN: Platform Bütünleşik EOTS Teknik İncelemesi
21 min read EN · TRTOYGUN'un nasıl gördüğünü, hedefi nasıl tanıyıp izlediğini, lazerle nasıl ölçüp işaretlediğini ve uçağın radar izini bozmadan tüm bunları nasıl yaptığını kademe kademe, sinyal zinciri ve alt sistemler ile anlatır.
- #ASELSAN
- #TOYGUN
- #EOTS
-
Fine-tuning an SLM on UAV combat doctrine
14 min read EN · TREnd-to-end notes from a $55 LoRA build on gpt-oss-20b: the data pipeline, the training run, the evaluation, and the surprise that the biggest win was teaching a reasoning model to stop reasoning.
- #LLMs
- #Fine-tuning
- #LoRA
-
The Silent Variable in Graph RAG: Why Corpus Language Matters More Than You Think
5 min read EN · TRChunk size, embeddings, re-rankers: the usual suspects. But the language of your corpus quietly shapes every layer of the pipeline, and reasoning models make it decisive.
- #Graph RAG
- #LLMs
- #Multilingual
Elsewhere
- Why is Monitoring Machine Learning Algorithms Difficult? (opens in a new tab)
- REST API'den Web Uygulamalarına: FastAPI, Flask ve Streamlit Rehberi (opens in a new tab)
- From Azure to Databricks: Data Analysis & Visualization Guide (opens in a new tab)
- ETL and Data Pipelines with Shell, Airflow and Kafka (opens in a new tab)
Publications
Certifications
MLOps Specialization — Duke University (2024)
Deep Learning Specialization — DeepLearning.AI (2022)
TensorFlow Developer — DeepLearning.AI (2022)
Fundamentals of Deep Learning — NVIDIA (2020)
SQL for Data Science — UC Davis (2024)
Generative AI for Everyone — DeepLearning.AI (2023)
Intro to Generative AI — Google Cloud (2023)
RPA Developer — LinkedIn Learning (2023)
Python 101 — IBM Cognitive Class (2020)