
Andrew Chappell developed advanced particle reconstruction features for the DUNE/dunereco and SBNSoftware/sbndcode repositories, focusing on low-energy neutrino event processing and particle flow analysis. He enhanced Pandora XML configurations and designed algorithms in C++ and Python to improve clustering, vertexing, and particle recovery, introducing dedicated workflows for high-density datasets. In SBNSoftware/sbndcode, Andrew implemented a merging algorithm for MIP-like stubs and shower cascades, recalibrating PFO scores to refine track and shower discrimination. His work integrated machine learning techniques, such as Boosted Decision Trees, and demonstrated strong workflow management, resulting in deeper analysis capabilities and more robust physics event reconstruction pipelines.
January 2026 (SBNSoftware/sbndcode): Focused on advancing particle flow reconstruction. Key feature delivered: a merging algorithm to combine MIP-like stubs with shower cascades from split primary electrons, and recalibration of PFO scores after potential electron merges to improve track/shower analysis. This work enhances reconstruction accuracy for electron-rich events and reduces misidentification between tracks and showers. The changes are implemented in SBNSoftware/sbndcode with commits 61db99a83b479d57faa6c3c059e50c2a03913a8c and 8a6de3ffb518311cad9c5d5c20ac1642fc9acae8. Major bugs fixed: none reported in this scope; focus has been on feature development and integration. Overall impact: improved particle flow reconstruction, better track/shower discrimination, and a foundation for more robust physics analyses. Technologies/skills demonstrated: algorithm design for particle flow, MIP-like stub/shower merging, PFO score recalibration, code integration into a large software repository, and version control discipline.
January 2026 (SBNSoftware/sbndcode): Focused on advancing particle flow reconstruction. Key feature delivered: a merging algorithm to combine MIP-like stubs with shower cascades from split primary electrons, and recalibration of PFO scores after potential electron merges to improve track/shower analysis. This work enhances reconstruction accuracy for electron-rich events and reduces misidentification between tracks and showers. The changes are implemented in SBNSoftware/sbndcode with commits 61db99a83b479d57faa6c3c059e50c2a03913a8c and 8a6de3ffb518311cad9c5d5c20ac1642fc9acae8. Major bugs fixed: none reported in this scope; focus has been on feature development and integration. Overall impact: improved particle flow reconstruction, better track/shower discrimination, and a foundation for more robust physics analyses. Technologies/skills demonstrated: algorithm design for particle flow, MIP-like stub/shower merging, PFO score recalibration, code integration into a large software repository, and version control discipline.
December 2024: Delivered production-network aware LowE workflow enhancements with Boosted Decision Trees (BDTs) integration for DUNE/dunereco. This update aligns the workflow with production environments and enhances analytical capabilities by integrating BDT-driven data processing steps and models.
December 2024: Delivered production-network aware LowE workflow enhancements with Boosted Decision Trees (BDTs) integration for DUNE/dunereco. This update aligns the workflow with production environments and enhances analytical capabilities by integrating BDT-driven data processing steps and models.
October 2024: DUNE/dunereco contributed enhancements to the Pandora XML configuration for low-energy neutrino detection and event reconstruction. This included new XML configurations and workflow tuning for low-energy processing, improving clustering, vertexing, and particle recovery accuracy. Added a dedicated Pandora low-energy workflow for HD datasets to support higher-density event reconstruction. Overall, feature-focused month with no critical bugs reported and strong alignment to business goals of improved sensitivity and data quality.
October 2024: DUNE/dunereco contributed enhancements to the Pandora XML configuration for low-energy neutrino detection and event reconstruction. This included new XML configurations and workflow tuning for low-energy processing, improving clustering, vertexing, and particle recovery accuracy. Added a dedicated Pandora low-energy workflow for HD datasets to support higher-density event reconstruction. Overall, feature-focused month with no critical bugs reported and strong alignment to business goals of improved sensitivity and data quality.

Overview of all repositories you've contributed to across your timeline