Raffaello Baluyot

Architected, designed, and developed components of the Autoliv Cloud Safety Platform (ACSP) to collect and save data from different research activities and partner companies. Current iteration of the ACSP is running and is processing more than 50GB of research data per day.

Designed and developed cross-platform SDK to enable researchers and partner companies to integrate with the ACSP. Development version of the Android App was developed and the fully developed iOS App was set to be published.

Researched electric scooter kinematics and video data to analyze driver behavior and driver interactions with other pedestrians as part of the e-Safe initiative. Research paper on handlebar interaction, pedestrian detection, and pedestrian position projection was published. A related project called Microvision was publicly funded to continue the study.

Reviewed, understood, and, validated different crash algorithms for performance, traceability and reproducibility for production applications. Concept validation report was published internally that point out the weaknesses of the crash algorithms. This led to further iteration of the algorithms and enlargement of the testing scope.

Collaborated with the Autoliv global infrastructure team on cloud system architectures and security. Improved networking, secret management, and data security of the existing services in Autoliv Research. Crucial platforms that require a security overhaul were being reworked.

Designed and developed a library and tool for processing high frequency vehicle logs. Data teams from other organization used the tool for creating their analysis and reports.

Designed and developed visual tools for monitoring vehicle faults and fault interactions. The tool highlighted 2 faults during electric truck development and was made available to all the system teams.

Designed and developed visual tools to show battery charging performance and behaviors. The tool was used by the charging team to ensure that the current charging performance meets the customer requirements.

Researched clustering applications to find different kinds of battery charging behaviors.

Analyzed the entire research process and requirements. A technical road-map for the project was developed.

Evaluated existing, and developed a new data gathering and evaluation strategy. Team was able to expand the number of raw data 25 times.

Mentored the researchers in development of tree (Canarium ovatum) classifier. A classifier meeting the target goals was developed, the feasibility of the research was validated, and the research moved to the next step.

Developed lessons and syllabi, and handled the following courses:
- Neural Networks
- Digital Image Processing
- Natural Language Processing

Handled the following courses:
- Advance Computer Architecture and Organization
- Design of Digital Systems and Computers

Redesigned, and redeveloped the alpha performance tool used for evaluation of trading signals. New alpha performance tool was 1.5 to 2 times faster, had a more modular code base, was the dependency of internal tools, and was used in the entire company.

Designed, and developed a tool to determine the contribution of individual models to the ensemble. Executives relied on the tool results to evaluate the performance of the current trading strategy.

Designed, and developed a tool to visualize the quality of structured data. The quantitative research and data operations teams used the tool to validate new and updated data streams.

Developed frameworks for creating different extract, transform, load (ETL) projects. The quantitative research and data operations teams used the framework to move research projects into production operations.

Researched and developed trading strategies from different concepts of liquidity. The trading strategy result was integrated as one of the part of the company trading strategy.

Researched and explored other strategies on equities, futures, ETFs, sectors, and risks. Quick research cycles, with initial results from 1-2 weeks of the research conceptualization.

Architected, designed, and developed a framework for deploying malware detection services. Machine learning models and analytical algorithms developed by the team were deployed to production using the framework.

Designed and developed a machine learning model for image file malware detection. The model was deployed in production as part of the larger malware detection pipeline.

Researched machine learning malware classifiers based on system logs and file metadata. The model was not stable as desired and the conclusion was that additional metadata were needed to improve stability.

Developed lessons and syllabi, and handled the following courses:
- Web Development
- Database Design

Architected, designed, and developed a chat bot framework supporting multiple chat platforms. All company chat bot offerings were ported or were created using the library.

Researched, and developed a machine learning model for GUI classification. The model showed feasibility and a vision-based automation product development started.

Researched NLP models on intent recognition, sentiment analysis, and phrase similarity. Developed English and Filipino models beat the performance of models of research partners.

Researched, and developed machine learning algorithm for tree cover monitoring. The contracts secured with government units and insurance firms moved the company focus towards remote sensing.

Architected, designed, and developed a cloud-based remote sensing GIS plug-in. The plugin was pitched to a cloud vendor who became a partner in selling the full solution.

Designed, and developed a backtesting and trading library for stock market trading research. The library was praised and used by internal researchers.

Design, and developed microservices for providing a credit scoring service to a client. The project reached the algorithm testing phase, processing thousands of credit applications per day.

Architected, designed, and developed FMI integration to CRAMAS ECU testing tool. CRAMAS Tool was able to integrate FMI compliant models from various vendors in simulation.

Designed, and developed virtual ECU simulation in CRAMAS. CRAMAS Tool and Virtual ECU prototypes showed feasibility and the project moved to next phase.

Maintained CRAMAS Auto Test Tool, a tool for creating test cases for automotive hardware simulation. CRAMAS Auto Test Tool received 10-20 bug fixes and performance improvements each quarter.

Developed a methodology for land cover classification using landsat. The methodology was adapted as team baseline for landsat processing.

Researched machine learning image classifiers for LiDAR and landsat. Resource mapping algorithm was developed and resource map for provinces was given to government units.

Represented the team in Phil-LiDAR 2 Luzon Cluster Colloquium. Presented the topic: Development of End-to-End Image Processing Pipeline for Resource Mapping.