Maik K.

Key Skills

Software

AWS SageMaker

Top Subject Matter

Medical AI / cardiology heart-sound analysis

Conversational AI

NER/entity extraction

Top Data Types

Audio

Text

Image

Document

Top Task Types

Diagnosis

Fine-tuning

Freelancer Overview

AI Engineer at ListenLabs (Heart Condition Detection System). Brings 12+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include AWS SageMaker, Hugging Face, and Transformers. Education includes Master of Science, Technological and Higher Education Institute of Hong Kong (2016) and Bachelor of Science, University of Hong Kong (2014). AI-training focus includes data types such as Audio, Medical, and DICOM and labeling workflows including Diagnosis and Fine-tuning.

Labeling Experience

AI Engineer at ListenLabs (Heart Condition Detection System)

AWS SageMakerAudioDiagnosis

Led development of an end-to-end heart condition detection system using labeled heart sound recordings and deep learning for medical diagnosis. Trained models on cleaned datasets derived from audio clips and improved performance by reducing false positives. Collaborated with domain experts to refine data models and analytics frameworks supporting supervised learning. • Trained deep neural networks on heart sound audio clips • Improved accuracy to 94.27% and reduced false positives by 20% • Tuned model components and paired networks with SVM/KNN/tree methods • Used Keras and TensorFlow for training and evaluation

2023 - 2025

AI Engineer at Google (Conversational AI, RAG, and entity extraction)

AudioFine-tuning

Built and fine-tuned multiple conversational AI and ML systems using labeled interaction data and model training workflows. Leveraged retrieval-augmented generation and orchestration (via LangChain) to improve response accuracy and reduce pipeline latency. Applied prompt-based learning and recursive reasoning techniques to refine extracted metadata entities. • Developed conversational AI systems using AWS and Nvidia NeMo • Leveraged LangChain for LLM orchestration • Optimized pipeline latency by ~35% and improved response accuracy • Applied few-shot, in-context learning, and recursive reasoning for entity extraction

2021 - 2023

Machine Learning Research Intern at Google

ImageFine-tuning

Conducted hands-on deep learning model development and training experiments, including CNN design and evaluation on labeled datasets. Implemented data preprocessing, augmentation, hyperparameter tuning, and validation to optimize model performance. Documented training procedures and research results for knowledge sharing. • Designed and trained a basic CNN for recognizing handwritten digits • Implemented data augmentation to improve robustness and generalization • Fine-tuned hyperparameters and validated the CNN for accuracy • Documented and presented development results and model effectiveness

2020 - 2020

Education

Work History