I operate at the intersection of high-performance localized AI compute, physical medical device fabrication, and sensitive clinical data compliance.
A scalable AI-to-Hardware framework allowing users to generate custom-fit CPAP masks using natural language prompts. Tested on myself as “Patient Zero,” this system processes user constraints, generates optimized 3D geometries, and undergoes an engineer-verified “Human-in-the-Loop” review before rapid biocompatible prototyping.
View Project Details: Generative AI CPAP Framework: Patient-Specific Anthropometric DesignArchitected and built a fully local, vendor-independent, sovereign AI research agent for an affiliated law firm (LM Studio · llama.cpp · self-hosted SearXNG · MCP tool-calling). It runs multi-hour unattended web and academic retrieval, returns structured, schema-validated output, flags un-scrapable sources for manual review, and keeps 100% of data on-device — preserving attorney–client privilege with no cloud, accounts, or vendor lock-in.
View Project Details: Vendor-Independent Local AI Research Agent (Law Firm Deployment)
An agentic Python pipeline bridging local distributed LLMs with FreeCAD’s API to automate complex 3D spatial modeling. It translates user-generated natural language prompts into executable parametric models, acting as the core mechanical middleware for rapid R&D.
View Project Details: Local AI CAD Assistant (FreeCAD)
Architected a private, high-performance AI ecosystem bridging a dual NVIDIA DGX Spark cluster with secure edge clients via Tailscale. This endpoint-agnostic infrastructure allows for manual orchestration of massive open-weight models (Qwen 3.5 122B, Llama 4) to provide localized, high-speed compute for complex engineering, research, and coding workflows.
View Project Details: Custom Local AI Network (DGX Spark Cluster)Applied research on making large language models run faster on local hardware, measured with reproducible benchmark tooling. A U.S. provisional patent application has been filed, so technical details are withheld for now.
View Project Details: LLM Inference Optimization (Patent Pending)Led a multi-institution clinical study (Penn State & University of Pittsburgh) quantifying the cognitive effects of sleep apnea. Developed a latency-free Python biometric suite using an AI-assisted development cycle to transition the project from localized pilot testing to a scalable, 45-subject remote experimental cognitive-behavioral study under formal IRB protocol.
View Project Details: Ongoing Research Project on Sleep ApneaThe custom reaction-time and working-memory battery behind my sleep apnea study: a 60,000+ line Python and PsychoPy application that participants run on their own Windows computers, built for sub-millisecond timing, accessibility, and verifiable data.
View Project Details: RCRFB: Open-Source Cognitive Testing Software
Pushed an ASUS ROG Phone 7 Ultimate to its absolute compute limits, securing a 3rd-place global AnTuTu benchmark score (1,591,570). By utilizing thermoelectric cooling, deep OS stripping, and aggressive thermal profiling, I transformed this flagship device not only into a benchmarking device but also into a sustained edge-compute node capable of running local AI models while maintaining daily-driver functionality.
View Project Details: Mobile Phone 3rd Global Benchmark record (2023)For BIOENG 1160 Senior Design my team is designing a medication management system for geriatric patients who struggle with medication compliance. It uses a rotating dispenser that drops one dose at a time into a cup that detects when the medication has been removed, along with a caregiver app.
View Project Details: Senior Design: Medication Management System (Team 11)
BIOE 1024 is medical product design course. Throughout this course I learned a variety of methods of prototyping, design, manufacturing, etc. I also had a term project that marked the first physical iteration of my AI generated custom fit CPAP mask.
View Project Details: BIOE 1024In Pitt’s SEA-PHAGES research course, I isolated and purified a new bacteriophage, Tricson, from soil, produced a high-titer lysate for genome sequencing, and annotated tail-cluster genes of the phage Starburst.
View Project Details: Phage Discovery and Genome Annotation (SEA-PHAGES)
For my high school Physics course I purchased an electric wheelbarrow conversion kit, added it to a wheelbarrow, added brakes, developed fabricated and added a riding tray unit, made custom battery packs and a custom battery holster on the RMEW.
View Project Details: Rideable Motorized Electric Wheelbarrow
For my Engineering Statics course, I designed and fabricated a standard truss bridge that exploited assignment constraints to achieve massive structural integrity. By laminating balsa wood with heavy JB Weld epoxy to create a custom composite material, the bridge successfully held over 150 lbs, ultimately breaking the testing equipment without sustaining a single point of structural failure.
View Project Details: Bridge ProjectKimbel Technologies2026 to present
I design, deploy, and maintain fully local AI systems for clients. One deployment I can share is a vendor-independent AI research agent for an affiliated law firm, which I continue to maintain.
See the deploymentChoose a skill to highlight where it shows up.
| Skill | AI CPAP Mask | Local AI Research Agent | AI CAD Assistant | Local AI Network | LLM Inference Optimization | Research Project | RCRFB Software | Phone OC project | Senior Design | BIOE1024 | SEA-PHAGES | Wheelbarrow Project | Bridge Project |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Software and AI | |||||||||||||
| No | No | No | No | No | No | No | No | No | |||||
| No | No | No | No | No | No | No | No | No | No | ||||
| No | No | No | No | No | No | No | No | No | |||||
| No | No | No | No | No | No | No | No | No | No | No | |||
| No | No | No | No | No | No | No | No | No | No | No | |||
| No | No | No | No | No | No | No | No | No | No | No | |||
| No | No | No | No | No | No | No | No | No | No | ||||
| No | No | No | No | No | No | No | No | No | No | No | |||
| Hardware and electrical | |||||||||||||
| No | No | No | No | No | No | No | No | ||||||
| No | No | No | No | No | No | No | No | No | No | No | No | ||
| No | No | No | No | No | No | No | No | No | No | No | |||
| No | No | No | No | No | No | No | No | No | No | No | No | ||
| Mechanical and fabrication | |||||||||||||
| No | No | No | No | No | No | No | No | No | No | ||||
| No | No | No | No | No | No | No | No | No | No | No | |||
| No | No | No | No | No | No | No | No | No | No | No | |||
| No | No | No | No | No | No | No | No | No | No | No | |||
| Research and medical | |||||||||||||
| No | No | No | No | No | No | No | No | No | No | No | No | ||
| No | No | No | No | No | No | No | No | No | No | No | No | ||
| No | No | No | No | No | No | No | No | No | No | No | No | ||
| No | No | No | No | No | No | No | No | No | No | No | |||
| No | No | No | No | No | No | No | |||||||
| No | No | No | No | No | No | No | No | No | No | No | No | ||
| No | No | No | No | No | No | No | No | No | No | No | |||
| No | No | No | No | No | No | No | No | No | No | No | |||
| Engineering process | |||||||||||||
| No | No | No | No | No | No | ||||||||
| No | No | No | No | No | No | No | No | No | No | No | No | No | |
| No | No | No | No | No | No | No | No | No | No | No | No | ||
| No | No | No | No | No | No | No | No | No | No | No | No | No | |
| No | No | No | No | No | No | No | |||||||
I operate at the intersection of high-performance localized AI compute, physical medical device fabrication, and sensitive clinical data compliance. My work bridges the gap between digital generative logic and tangible, FDA-aligned physical prototyping.
My primary expertise lies in Edge Compute and Local LLM Orchestration, where I architect multi-node enterprise compute ecosystems to orchestrate massive open-weight models (122B+ parameters). I construct highly secure, endpoint-agnostic infrastructure optimized for absolute data sovereignty, specifically designed to eliminate data capture by device manufacturers. Additionally, I specialize in Medical Device Prototyping (taking digital CAD models through FDA design controls and ISO 10993 biocompatible material selection) and Hardware Thermal Profiling, having secured a 3rd-place global AnTuTu benchmark record by deploying custom thermoelectric cooling arrays for sustained local AI compute.