Joel Sprunger

About Joel Sprunger

Experienced Staff Engineer with a demonstrated history of innovation in signal processing, machine learning, and Generative AI. Skilled in Python, JavaScript, Matlab, C/C++, Verilog/VHDL, and LabVIEW. Strong engineering professional with a Master of Science (MS) focused in Electrical & Computer Engineering (Signal Processing) from Portland State University.

Experience

1. Full Stack Ai Engineer

   Lender Toolkit

   Jun 2023 — Present · US

   Agent Platform• Guidelines Agent: Built a LangGraph agent over Postgres with pgvector that can handle complex queries with accurate information and citation; also designed additional specialized agents for text extraction, loan data, and general‑purpose assistance that build on the same agentic patterns.• Guideline Updater Service: Architected and implemented an end‑to‑end ETL pipeline that crawls and normalizes official guideline sources using Firecrawl, custom parsers/processors, and a resilient async pipeline with Redis queues, retry mechanisms, and CloudWatch monitoring.• TypeScript/React Frontend: Web app via a modern streaming chat experience with conversation management, user settings, model/agent selection, streaming tool status, markdown rendering with LaTeX support, and structured visualization rendering.• Leadership: Mentored and directed 2 interns and 2 contractors (requirements, ADO/GitHub planning, code and architecture reviews), effectively acting as the technical lead for the guidelines and agent stacks.Additional Features• AI Filter Selection: Use natural language to filter data displayed in the web app• Loan Summarizer: Steamed summary sections, supported by a recursive JSON scrubber/splitter (contributed to LangChain).• AI Profile Images: Take profile picture -> vision model analysis -> HuggingFace Avatar Generation. • Tetris Game: Created a production‑quality React Tetris game based on the company logo Architecture LeadershipServed as lead technical/architectural reviewer for a complex pipeline agent delivered by an external consulting partner, redesigning it into a transparent, pandas‑style agentic workflow with explicit tools and iterative refinement instead of a rigid one‑shot pipeline, resulting in a more reliable design which is less prone to hallucination and more efficient with tokens.

Education

• Portland State University

  Master of Science (MS), Electrical & Computer Engineering

  2015 — 2018

• Oregon Institute of Technology

  BSEET, Electrical Engineering/Signal Processing

  2008 — 2011

Skills

• Uflex
• Particle Filters
• State Space Tracking
• Engineering
• Fpga
• Mathematics
• Verilog
• Kalman Filtering
• Vhdl
• Labview
• Signal Processing
• Matlab
• Ltspice
• Vivado
• Image Processing
• Ate
• Microsoft Office
• Array Processing
• Automation
• Pdm
• Xilinx Ise