Haruka Inoba

- Built backend components to process and serve 10K+ real-time time series records for a web dashboard - Integrated Apache Pinot API to enable low-latency data access between the service and analytics datastore - Optimized SQL queries and KPI computations within the service layer to improve dashboard responsiveness and performance

- Developed a data conversion pipeline in Python to transform XML-formatted historical data stored in SQL into JSON for Elasticsearch indexing - Implemented backend data transformation logic to enable graph-based exploration of Roman history in Neo4j

Small ML projects: - Saliency prediction Team project: - Empirical Research between Trustworthiness and robot's Voice in Human-computer interaction Master's thesis: - Compensation for Imbalanced In-the-wild Emotional Speech Dataset for Emotional Speech Synthesis - Implemented a network applying the data augmentation technique and Integrated the trained network into the Emotional Speech Synthesis model (Generative AI) -Conducted experiments, and analyzed results to derive key insights

Bachelor’s thesis: - Extraction of Car Interior Acoustic Features Related to Auditory Impression Using Artificial Intelligence - Role: Collecting results through experimental runs, analyzing and reporting findings to supervisors and researchers at the collaborating company, - Technologies used: MATLAB, machine learning techniques, and various preprocessing methods for data processing and analysis