Headshot of Jack P. DeMarinis

Jack P. DeMarinis

Education

Master of Science: Electrical Engineering

University of Rhode Island, Kingston, RI - Expected May 2026

Accelerated B.S./M.S. (ABM) Program

GPA: 4.00 / 4.00

Bachelor of Science: Computer Engineering

University of Rhode Island, Kingston, RI - May 2025

Minor: Mathematics

GPA: 3.90 / 4.00

Technical Skills

GenAI / LLM

  • OpenAI API
  • Gemini API
  • LangChain
  • Agentic architectures
  • MCP servers
  • Retrieval-augmented generation (RAG)
  • Embeddings
  • Prompt engineering
  • Local small language models

Languages

  • Python
  • C++
  • C
  • Bash
  • JavaScript
  • HTML/CSS
  • MIPS Assembly
  • LC-3

Tools & Design

  • Visual Studio Code
  • Unity
  • Fusion 360
  • AutoCAD
  • MATLAB
  • Multisim
  • Mathcad
  • VHDL
  • OpenMV IDE
  • Git / GitHub
  • ROS / ROS2

ML / AI

  • PyTorch
  • Hugging Face Transformers
  • Hugging Face Datasets

Backend / Systems

  • Linux
  • Docker
  • Google Cloud Console (GCP)
  • PostgreSQL

Experience

Graduate Research Assistant

University of Rhode Island, Kingston, RI

  • Designed agentic, multi-step decision pipelines with tool calling and structured outputs to improve reliability across complex workflows.
  • Built reusable MCP-style tool interfaces connecting LLMs to external services and internal utilities.
  • Implemented simulation environments for swarm robotics in 2D, 3D, and VR using Pygame and Unity and latency-sensitive behaviors.
  • Developed reproducible Linux and Docker workflows to support consistent builds and deployments across machines.
  • Implemented and maintained a PostgreSQL database to support reliable data storage and fast query access.

Undergraduate Research Assistant

University of Rhode Island, Kingston, RI

  • Built LLM features using RAG, improving factual grounding across evaluation prompts.
  • Integrated LLM capabilities using OpenAI's API in combination with GCP, connecting to Gmail and Google Calendar and implementing Google OAuth for secure sign-in.
  • Developed and shipped backend services supporting AI workflows handling hundreds of requests per day.
  • Created evaluation scripts to compare LLM variants using consistent test sets and measurable latency, cost, and output-quality criteria.

Computer Engineering Intern

Electro Standards Laboratories, Cranston, RI

  • Developed Python and C software on embedded Linux platforms, improving system reliability through repeatable test routines.
  • Diagnosed and resolved electrical and software issues using structured debugging, instrumentation, and logging.
  • Produced clear technical documentation to improve maintainability and team handoff.

Software Engineering Intern

IGT, West Greenwich, RI

  • Resolved Linux system issues using low-level command-line debugging and root-cause analysis.
  • Developed Bash, C, and C++ code for device-level integrations and internal tooling.
  • Implemented OCR pipelines and validation checks to improve system accuracy and robustness.

Projects

AI Meeting Assistant

  • Shipped a production full-stack app and deployment workflow on Railway with a PostgreSQL-backed data model for durable meeting artifacts.
  • Designed an LLM workflow with schema-first JSON outputs plus validation to standardize summaries and action items for downstream integrations.
  • Built reliability into the pipeline (retries, idempotent writes, explicit failure states) to prevent partial processing and inconsistent data.
  • Created reusable prompts/templates and documentation to make the workflow repeatable for new users and future features.
Visit Meeting Muncher

Senior Capstone: Robotic Assembly & Inspection

  • Built a modular automation workcell for through-hole PCBA assembly and inspection.
  • Designed end effectors, fixtures, and feeder hardware in CAD; iterated via rapid prototyping.
  • Integrated OpenMV vision for pose/orientation detection and UART/serial signaling to control code.
  • Wrote documentation and handoff notes (design decisions, layout, troubleshooting).
  • Presented the final system to a large audience, covering results and design tradeoffs.
View Project Report

Agentic RAG Chatbot

  • Built a multi-agent retrieval-augmented generation system for structured reasoning.
  • Implemented embedding pipelines and FAISS-based vector search to reduce hallucinations and improve response quality.
  • Deployed the system as a Dockerized Flask API designed for iterative testing and evaluation.
Try the Chatbot

Honors & Activities

Contact

Open to new opportunities and collaboration. Reach out anytime.