Alaettin UÇAN

Alaettin UÇAN

Phd @ CS, Text Mining, Deep Learning, Optimization, Pretrained Language Models, Fullstack Machine Learning Engineer

Portfolio

Project-26-NanoLoom

NanoLoom aims to revolutionize drug discovery and molecular diagnostics by developing an imaging-based microfluidic platform that analyzes DNA-protein interactions at the single-molecule level. It combines nanoimprint-based microfluidic chips for DNA stretching and imaging, TIRF microscopy for visualizing molecular dynamics, and AI-powered software tools for quantitative analysis of protein binding and drug effects, providing a high-throughput, cost-efficient alternative to existing biochemical assays. Funded by the Eurostars-3 programme by Horizon Europe. (2025-Present)

Project-25-ExerNeck: AI-Driven Mobile Tele-Rehabilitation & Gamification

Directing the R&D strategy for a sensor-less, mobile tele-rehabilitation platform designed to treat cervical musculoskeletal disorders caused by sedentary lifestyles. The solution leverages real-time computer vision (MediaPipe) to analyze patient movements via standard smartphone cameras, and integrates Promptable Game Models to dynamically generate personalized gamification content, boosting patient adherence. Supported by the TUBITAK 1501 program. (2025-Present)

Project-23-HIVEMIND: LLM-Based Multi-Agent Software Engineering Framework

Contributing to a Horizon Europe (CL4) initiative designed to revolutionize the Software Development Lifecycle through Agentic AI. The project introduces an adaptive LLM-based multi-agent framework that facilitates collaboration between human developers and autonomous AI agents, assigning distinct, role-based personas to AI to advance responsible, human-centric software engineering while maintaining ethical oversight and code quality. (2025-Present)

Project-22-Sym2Clinic: Generative AI-based Patient Information and Guidance System

Directed the R&D strategy for a Generative AI platform designed to revolutionize the pre- and post-examination patient journey. The system utilizes Large Language Models to analyze natural language patient complaints, performing automated clinical triage to route patients to the appropriate specialty and assess eligibility for telemedicine, optimizing scheduling and follow-ups. Supported by the TUBITAK 1501 program. (2024-Present)

Project-21-RAD-THERA: AI-Driven Precision Radiotherapy & Dosimetry Platform

Directed the R&D strategy for a deep-learning-based radiotherapy planning system designed to automate precise segmentation of Target Tumor Volumes and Organs-at-Risk. Leveraging multi-modal imaging (CT/MRI), the platform optimizes radiation dose distribution to maximize therapeutic efficacy while minimizing toxicity, reducing manual contouring workload and inter-observer variability. Supported by the TUBITAK 1501 program. (2024-Present)

Project-20-My Health Coach: AI-Powered Personalized Health Notification and Risk Assessment System

Spearheaded the R&D strategy to transform legacy, passive Personal Health Records (PHR) into a proactive, AI-driven health management ecosystem. The platform leverages Graph Machine Learning to model patient-disease relationships and family history for chronic disease risk stratification, and Reinforcement Learning to optimize the timing and content of personalized health interventions. Supported by the TUBITAK 1501 program. (2024-Present)

Project-19-Fallower: Real-Time UWB Based Human Fall Detection and Tracking System

Directed the R&D and AI strategy for Fallower, a non-intrusive safety monitoring system utilizing Ultra-Wideband (UWB) radar technology. Engineered a privacy-preserving alternative to camera-based or wearable solutions, employing machine learning algorithms to classify human movement patterns and detect fall anomalies with high precision, triggering instant alerts to caregivers. Supported by the TUBITAK 1501 program. (2024-Present)

Project-18-DMEK-Track: AI-Driven Corneal Transplant Monitoring System

Led the development of a clinical decision support system designed to optimize post-operative follow-up for Descemet Membrane Endothelial Keratoplasty (DMEK). Deployed advanced deep learning architectures (Operational Neural Networks and CNNs) to analyze Anterior Segment OCT imagery, automatically assessing graft health and triaging patients for advanced confocal microscopy. Supported by the TUBITAK 1507 program. (2024-2025)

Project-17-Autononym: AI-Driven Data Privacy & Compliance Platform

Directed the R&D strategy for a comprehensive automated anonymization solution designed to secure sensitive data across high-regulation sectors (Health, Banking, Public). Utilizing a multi-modal AI pipeline (advanced NLP and image processing), the platform automatically classifies and de-identifies PII within unstructured text, tabular data, and images, balancing GDPR/KVKK compliance with data utility. Supported by the TUBITAK 1507 program. (2023-2025)

Project-16-Mobithera: Remote Physiotherapy Platform

Spearheaded the R&D and AI strategy for Mobithera, Turkey’s first MDR-certified Digital Therapeutic (DTx) for remote physiotherapy. Engineered highly optimized, proprietary Edge-AI Human Pose Estimation models running locally on mobile devices for real-time, privacy-centric (HIPAA/GDPR) biofeedback. The platform leverages Unity-based gamification to drive patient adherence and has achieved FDA Approval, EU MDR Certification, and NHS ORCHA validation. Supported by the TUBITAK 1501 program. (2023-Present)

Project-15-Predis: Anomaly Detection, Early Warning and Forecast System

Directed the R&D and AI strategy for an advanced predictive analytics platform designed to safeguard the pharmaceutical supply chain. The system leverages Time-Series Foundation Models (Amazon Chronos) for high-precision forecasting and unsupervised learning (Local Outlier Factor) to detect stock anomalies in real-time, integrated into GIS-enabled dashboards for health authorities. Supported by the TUBITAK 1501 program. (2023-2024)

Project-14-Cloud Based Health Information Hub & Exchange

Spearheaded the Machine Learning and Analytics architecture for a major R&D initiative to transform legacy national-scale health solutions (EHR, Drug Tracking, e-Prescription) into a globally scalable, cloud-native ecosystem. Engineered a resilient microservices infrastructure orchestrated via Kubernetes and Apache Kafka, integrating a centralized Data Warehouse for unified analytics. Supported by the TUBITAK 1501 program. (2023-2024)

Project-13-AI-Driven Robotic Chemotherapy Preparation System

Directed the R&D and AI strategy for an indigenous robotic system designed to automate high-risk chemotherapy drug preparation, mitigating critical safety risks including contamination and foaming. The system ensures end-to-end medication traceability by integrating directly with Turkey’s national health ecosystems (ITS and e-Prescription). Supported by the TUBITAK 1501 program. (2023)

Project-12-AISym4MED: Synthetic Data Generation & Privacy Platform

Contributing to a major EU-funded initiative (Horizon Europe) designed to address data scarcity and privacy bottlenecks in medical AI. The project is developing a trustworthy platform that generates high-fidelity synthetic medical datasets via controlled data synthesis, integrating novel anonymization pipelines and attribute-based privacy measures to provide a compliant, scalable data ecosystem for healthcare data engineers and practitioners. (2023-Present)

Project-11-AI-Driven Automated Assessment Generation Platform (OSYM)

Spearheaded the development of an automated question generation system for OSYM (Student Selection and Placement Center). The project utilized a sophisticated NLP pipeline, fine-tuning BERT and T5 architectures for context-aware question generation and paraphrasing, employing semantic word embeddings to engineer plausible multiple-choice distractors, and was delivered as a scalable Machine-Learning-as-a-Service API (PyTorch, Transformers, Django, PostgreSQL). (2021-2022)

Project-10-YZYap: AI-Enabled Educational Robotics Platform

Engineered an interactive educational robotics platform designed to democratize AI and programming concepts for beginners, supported by the TUBITAK 1501 R&D program. The system features a versatile dual-mode coding environment (Scratch and standard languages) integrated with hardware sensors (camera, microphone) for multimodal human-robot interaction, plus a modular AI interface for sentiment analysis, object tracking, and image recognition. (2020-2021)

Project-09-Pioneering Turkish Emotion Analysis with LLMs

Executed one of the first studies to fine-tune pre-trained Large Language Models specifically for Turkish emotion analysis using curated, domain-specific datasets. By leveraging novel transfer learning techniques, the research achieved state-of-the-art accuracy scores, surpassing traditional baselines. This work resulted in 3 published papers and established a new benchmark methodology for Turkish computational linguistics. (2017-2020)

Project-08-TCMB Banknote Production Tracking System

Developed a high-security monitoring platform for the Central Bank of Turkey’s Banknote Printing Plant to ensure end-to-end traceability of machinery, personnel, and sensitive raw materials. The system optimized production efficiency through real-time statistical analysis while safeguarding operations through an automated anomaly-detection and alert mechanism. (2017-2021)

Project-07-Automated Media Intelligence System for Turkish General Staff

Designed and deployed a mission-critical media monitoring platform for the Turkish General Staff to automate the digitization and analysis of daily print media. Leveraging the Abbyy FineReader API for advanced OCR processing and C#.Net for a robust backend architecture, the system automatically detected and highlighted sensitive keywords within massive text data, streamlining the workflow for operators to compile daily intelligence briefings. (2017-2021)

Project-06-Ford Otosan Production Line Laser Automation System

Engineered a mission-critical laser marking automation solution for the Ford Sakarya Factory by reverse-engineering proprietary production protocols to synchronize real-time vehicle data with the manufacturing process. Built on a hybrid C++ and C#.Net architecture, the system orchestrates the precise operation of the laser unit and peripheral electromechanical components, delivering a high-stability solution that has operated continuously on the production line for years with zero downtime. (2017)

Project-05-NLP Tooling for Linguistic Research

Developed a specialized n-Gram extraction and text mining engine for the Hacettepe University Department of Linguistics. The tool utilized Python (NLTK, Zemberek) to process complex Turkish morphological structures, enabling advanced academic research on large text corpora. (2017)

Project-04-Cross-Lingual Sentiment Lexicon Generation Framework

Addressed the scarcity of annotated data in low-resource languages by developing an automated framework to generate Turkish sentiment lexicons from English resources. Implemented a novel triple unification process to map sense-level polarities to word-level scores and employed a multi-algorithm bilingual translation strategy to resolve semantic ambiguity. Supported by the TUBITAK 1001 research grant. (2016-2017)

Project-03-Zigbee-Based IoT Ecosystem & Smart Home Platform

Architected and developed a comprehensive smart home ecosystem for Gesislab Elektronik, supported by the TUBITAK 1501 program. Designed a centralized, web-based command hub comparable to industry leaders like Google Home, enabling seamless wireless device orchestration via the Zigbee protocol. Managed a heterogeneous technology stack, integrating low-level firmware (C, C++) with dynamic backend services (Python, C#.Net) and a hybrid database layer (PostgreSQL, NoSQL). (2015-2018)

Project-02-Market Star Metastock-third-party

Engineered a third-party financial analysis plugin tailored for the Metastock trading platform using a hybrid C++ (DLL) and C#.Net architecture. Beyond the core analytical engine, designed and deployed a complete commercial infrastructure, including a secure licensing system, integration with a virtual POS payment gateway, and automated installation packages. Also delivered comprehensive user support modules and documentation to ensure seamless onboarding. (2013)

Project-01-Kardelen Software-HIS

Took part in the design and implementation of several modules of a nationwide Hospital Information Management System (HIMS) deployed in over 200 hospitals across Turkey. Engineered scalable, n-tier software modules for Surgery Management, Clinical Operations, Inventory and Accounting using Object-Oriented design principles, supporting thousands of concurrent users. (2008-2010)