Serving in a cross-domain capacity at a dynamic startup, responsibilities span a wide spectrum, from code development to client interactions and data engineering endeavors. The primary task involves the characterization and acceleration of benchmarking codes for a unique computational system. Through the adaptation of C++, OpenMP, and MPI-accelerated benchmarking, the versatility of the product across multiple verticals, including bioscience, encryption, finance, and particle simulation, was showcased.Introduced automated workflows, streamlining processes and enhancing efficiency across projects. Insights from regular interactions with the business development team and clients shaped the product's direction, ensuring alignment with market needs. A significant contribution was made as an early developer to a greenfield project, focused on estimating and modeling the new generation of products, now reaching maturity and dispatched to clients.Languages such as C++, C, and Python, along with OpenMP, MPI, and bash, became instrumental. Internal tools for low-level language deconstruction, LLVM IR, and various acceleration frameworks were navigated. Emphasis on data engineering strategies, coupled with automation and metric analysis, ensured operational efficiency and data integrity.

Full-Stack Development & Advanced Data Engineering for Autonomous Heavy Machinery SystemsSole developer engaged in full-stack development and advanced data engineering tailored for autonomous heavy machinery systems. Backend solutions, crafted in Python and C, seamlessly integrated WebSocket, DDS protocols, SQLite, and InfluxDB, laying a foundation for efficient real-time data processing.On the front end, a dynamic interface was developed using Angular, resulting in a versatile Progressive Web Application (PWA). Features of this PWA included user-customizable dashboards, real-time data visualization with minimal latency, and integrations like LIDAR visualizations and HUD developed with Three.js for real-time GPS-based tracking.Efforts in data engineering prioritized high-frequency strategies, focusing on data ingestion, processing, and storage, ensuring data integrity. The architecture and management of a data lake were spearheaded, encapsulating agile data collection and retrieval to meet dynamic system demands.In the realm of Tools & DevOps, Golang was utilized for versatile data ingestion tools, and Jenkins orchestrated CI/CD pipelines, ensuring efficient development cycles.

Automated Segmentation of Missing Wedge Artifacts in Cryo-electron Tomography Maps by Shape-Constrained Deconvolution. Extended research and developed software packages for several novel methodologies on shape-constrained deconvolution of 3D Cryo-ET intensity maps of Actin filament within Stereocilia in Python and C. Methods allotted for super-resolution tracing of filaments and successful removal of “Missing Wedge” artefacts. Inverse problems both linear and non-linear in nature were investigated as well as specialized optimization methods based off ADMM and RL Deconvolution. Successful use of GPU acceleration allows for a several magnitude speedup of computation time. Work presented at BPS 2020 and published in JCIM.

Member of an Agile team responsible for the development of several large enterprise level .NET / SQL products, with a focus on Library Science applications. Curated several legacy codebases and developed rapid solutions and features within an elevated rollout tempo. Utilized Azure DevOps with Git for product management and team coordination. Investigated quality control automation for integration testing purposes. Designed, developed, and overhauled Microsoft SQL databases for large-scale features and performance optimizations. Investigated and developed components for CI/CD automated installation and UI testing within Azure DevOps. Experience in both developing and interfacing with RESTful APIs and web pages within ASP.NET. Developed features for several legacy applications with C# and WinForms. Developed interfacing addons in LUA. Maintained and developed several projects which utilized both microservice and monolithic architectures.

Consisting primarily of two separate projects - a Jargon-based NLP/speech-to-text framework/application and investigations into real-time prognosis/EOL prediction model utilizing FPGA co-processors.1) Jargon Based Speech to Text Application.Development of tool for the transcription of domain specific knowledge local to the individual subfields present at NASA Langley. Included the use of Kaldi libraries for customized language corpus training and deployment, server based infrastructure with worker mechanisms for serving multiple instances, and a web based front end offering real time transcripts to any potential user. Further work underway for the disambiguation and linking of NASA acronyms within the application in collaboration with a team from Virginia Tech.2) Field Programmable Gate Array Co-processors for the Efficient Implementation of Data Driven Prognosis.Through the utilization of AWS F1 instances and respective FPGA Xilinx co-processors, developing, implementing, and deploying predictive models. Exploration in neural network FPGA deployment, network integer quantization, and customized operation linking. Evaluation of computational acceleration, power efficiency, reduction in mass and volume, and real time capabilities of systolic model implementations for edge computing and field applications.

In Situ Characterization for Additive Manufacturing: Utilizing high performance computing (HPC).Developed a framework to succeed in encapsulation of a near-closed loop system which offers capabilities in on-the-fly modification of structurally functional, topologically optimized components within a 3D printed system. Sensor fusion and data processing enabled through utilization of high volume real-time data processing, parallel computing methods, and datastructure considerations utilizing interfacing with high performance GPU and CPU libraries. Work presented at 2017 SIAM Annual Conference.

Developed a number of key capabilities, including an instance-based retrieval for image similarity (image searching) and a PubMed harvester for Watson Data Analytics ingestion.1) An instance-based retrieval for image similarity - utilizes deep neural networks and locality hashing (k nearest neighbors) to provide server based retrieval of images based on image similarity and/or facial recognition/features. Scaled the framework to production capabilities (several hundreds of thousands of images),2) A scheduled automated data crawler for PubMed that allowed for periodic data ingestion within a Watson Data Analytics platform.Aided in several projects with a number of auxiliary roles.

Part of a Pre-seed collection of founders exploring potential product avenues in the data science domain. Produced several full-featured products for candidate partners.

Curator and digital archive engineering -Designed, developed, implemented procedures, and programmed software for the extraction and digitization of over 3 decades of flight test and advanced avionics data from lidars, GPS, TCAS, ADS-B, and a variety of aircraft state-sensors. Recordings represent the development and testing of almost every major commercial avionic system/technology since 1985.

Utilized test equipment, schematics, and technical manuals to diagnose system malfunctions to the component level. Performed troubleshooting, repair, installation/removal, alignment, testing, wiring, and general preventative maintenance on assorted electronic and electric systems.