Kipso
Contact me

Summary

Engineer with hands-on experience building backend systems, AI-powered document processing pipelines, and self-hosted LLM inference infrastructure. Currently pursuing a Master's in Computer Science (OMSCS) at Georgia Tech with graduate-level coursework in Artificial Intelligence and Machine Learning. Background spans industrial automation (PLC/SCADA) at SCIO Automation/CEC Controls and embedded systems and defense hardware testing at BAE Systems.

Education

Georgia Institute of Technology

Jan 2024 – May 2028

M.S. Computer Science

AI (CS6601), Knowledge-Based AI (CS7637), Deep Learning (CS8001/NVIDIA DLI), Computer Vision (CS6476), Machine Learning for Trading (CS7646, ongoing).

University of New Hampshire

May 2019

B.S. Electrical Engineering

University of Massachusetts Lowell

Post-graduate work in Radar Systems

Certifications

  • NVIDIA Deep Learning Institute – Fundamentals of Deep Learning (2025)
  • NVIDIA DLI – Applications of AI for Anomaly Detection (2025)
  • NVIDIA DLI – Building AI-Based Cybersecurity Pipelines (2025)
  • NVIDIA DLI – Generative AI with Diffusion Models (2025)

Experience

Controls Engineer · SCIO Automation / CEC Controls

08/2022 – Present

  • Programs and maintains PLC systems using Allen-Bradley platforms (CompactLogix / ControlLogix) and SCADA systems including VTScada for real-time monitoring and control.
  • Supports system commissioning, debugging, and field integration across multiple project sites; interprets electrical drawings and control schematics.
  • Developed and deployed Watchtower, an internal vulnerability tracking system to monitor CISA advisories and firmware risks across customer panels and industrial control systems, supporting automated ingest, filtering, and long-term tracking of firmware/OS-level exposures.
  • Built and integrated a self-hosted DeepSeek AI server on a high-performance AMD EPYC platform to support local document querying and LLM-powered engineering insights.
  • Designed a multi-stage AI pipeline for interpreting industrial specification PDFs and addenda, generating structured engineering outputs (requirements lists, out-of-scope items, IO mappings, and page references).
  • Implemented PDF parsing, prioritization logic, and structured JSON/Markdown outputs using Python, PyMuPDF, FastAPI, and PostgreSQL/SQLite.
  • Integrated local LLM inference (Qwen/DeepSeek) and vector search (LocalRecall) to support RAG-based document querying and reasoning.
  • Collaborated with senior engineers and IT staff to ensure alignment with compliance, uptime, and security expectations across multiple client sites.

Electrical Engineer 2 · BAE Systems

Jul 2021 – Aug 2022

  • Supported design and analysis of electrical systems, reviewed schematics and hardware configurations for compliance with engineering standards.
  • Participated in testing, validation, and troubleshooting of electronic systems; worked with cross-functional teams on system integration.
  • Developed automation scripts in MATLAB and VBA to streamline engineering workflows and reduce manual data processing.
  • Reverse-engineered and debugged legacy hardware modules, identifying root causes of system failures.

Electrical Engineer 1 · BAE Systems

Jun 2019 – Jul 2021

  • Collected and analyzed test data to validate system performance under demanding operational conditions for advanced defense projects.
  • Created Excel VBA macros to simplify data extraction from large engineering datasets, improving efficiency for cross-functional teams.
  • Reviewed 100+ technical documents and schematics for compliance, submitting and supporting detailed engineering change requests (ECRs).
  • Authored repeatable procedures for firmware loading and system sanitization, enabling consistent handoff to lab technicians.

Selected Projects

Personal RAG & LLM Knowledge Base

Personal · Python, FastAPI, SQLite, Ollama, Qwen3, DeBERTa, BM25, HyDE, PyMuPDF

Local-first RAG system with no cloud dependency. A DeBERTa-based 11-intent classifier routes each query to RAG, web search, conversation history, or direct LLM answer. Retrieval uses HyDE + BM25/RRF hybrid scoring with MMR diversity. Includes an Extraction Studio that digests large documents into structured markdown reports via a scan → rehydrate → synthesize pipeline. Web UI has SSE streaming, collection scoping, conversation history, and a document library browser. Achieves 90% accuracy on a 20-question golden eval.

Applied AI & ML — OMSCS Coursework

Academic · Python · CS6601, CS7637, CS6476, CS7646

Four completed graduate courses: implemented search algorithms (BFS/A*/minimax), probabilistic models (Bayesian nets, HMMs, GMMs), a full CV pipeline (HOG+SVM detector, Kalman/Particle tracking), and a quantitative trading system with Q-learning/Dyna RL agent.

Aura – Personal AI Assistant

Personal · FastAPI, SQLite/PostgreSQL, FCM, OpenAI APIs

Personalized assistant for task tracking, short/long-term goals, push notifications, and daily message generation. Modular task/goal schema with FCM push logic.

Technical Skills

Languages & Web

  • Python, JavaScript, TypeScript
  • Node.js, React, Next.js, HTML, CSS
  • FastAPI, Flask, Uvicorn
  • Basic: C, C++, C#, Java

AI / ML & LLM Systems

  • Local LLM inference & self-hosted infra
  • Embeddings & RAG pipelines
  • Vector search, OpenAI API
  • Prompt engineering, PyMuPDF

Data, Tools & Engineering

  • SQL, PostgreSQL, SQLite
  • MATLAB, VBA, Excel Macros
  • Git, Linux, VS Code, PyCharm
  • PLCs, HMIs, SCADA, Lab instrumentation