Pruitt Colon

Full Stack AI Engineer

Casa Grande, AZ PruittColon@gmail.com (207) 337-0340
Full Stack AI Engineer with Master's in Software Engineering (AI focus). Built a privacy-first AI platform running entirely offline - data never leaves your device. Deployed 22 ML engines on a single GPU via novel semaphore design. I bring a reliability-first mindset to ML engineering, creating documented processes that save six figures and automation that becomes the team standard.
10+ years focused on operational efficiency across telecom, tech, and AI. I build systems that reduce manual work and create repeatable processes.
Focus on Operational Efficiency

Streamlining workflows has been central to every role I've held.

At SubCom, I became the go-to resource for integration procedures - studying timing data to identify bottlenecks and training teams on improved methods.

At BTI Solutions working in Samsung's office on unreleased devices, I identified repetitive tasks and built automation scripts that saved hours of manual work daily.

Why AI?

LLMs offered a way to accelerate the process improvements I'd been doing manually.

AI is a tool, not magic - no different from any other tool in a professional's toolkit. I use it to learn faster, validate ideas quicker, and automate repetitive tasks.

The judgment calls, architecture decisions, and security considerations remain human responsibilities. The value is in knowing when and how to apply it.

What I Built

Vox Amelior is a production-grade AI platform I built from scratch.

It handles real tasks: transcribing meetings, tracking conversations, analyzing data, and providing an AI assistant that understands context.

The platform grew into 12 microservices, 22 ML engines, and 332K+ lines across 7 coding languages - all running on a single consumer GPU with enterprise-grade security. See the full architecture

Core Competencies

AI/ML Deep Learning, Neural Networks, NLP, PyTorch, Generative AI, Prompt Engineering, Real-Time ASR (NVIDIA Parakeet), Speaker Diarization, RAG (FAISS + MiniLM), LLM Integration, Emotion Analysis, Causal Inference
MLOps & Infra GPU Optimization & Semaphores, Docker, Kubernetes, Istio Service Mesh, CUDA, Redis, PostgreSQL, Prometheus, OpenTelemetry, CI/CD, HPA Autoscaling
Software Eng Python, FastAPI, REST APIs, WebSocket, Microservices, RBAC, JWT/mTLS Authentication, Playwright E2E Testing, Event-Driven Design
Cloud & DevOps Linux, AWS (S3, Lambda, EC2), Git, Agile, Infrastructure as Code, Flutter, Dart, Deepgram STT, n8n Workflow Automation
Vox Amelior - Cognitive AI Orchestration Platform Featured
  • Built 22 specialized ML engines (AutoML, Monte Carlo, Granger causality, time-series forecasting, Graph Neural Networks) with System 2 cognitive architecture for mathematical validation of LLM outputs
  • Solved GPU contention via Redis-backed semaphore design enabling Gemma 3-4B + NVIDIA Parakeet ASR to coexist within 6GB VRAM with preemptive context-switching
  • Architected zero-trust security: mTLS authentication, encrypted vector stores (FAISS + SQLCipher), rotating JWTs, and comprehensive threat modeling
  • Deployed 13 containerized microservices via Docker/Kubernetes (Istio, HPA autoscaling, GPU time-slicing) with FastAPI gateway implementing RBAC and OpenTelemetry tracing
  • Built Flutter mobile app with Deepgram real-time STT and Fiserv DNA banking API integration for enterprise account/transaction management

Professional Experience

Network Engineer

BTI Solutions | San Diego, CA | June 2024 - June 2025

  • Created ADB automation scripts to streamline morning workflows, methodically deploying across every office workstation and improving testing deadline adherence
  • Executed Samsung 5G/LTE site validation while maintaining strict uptime SLAs for Tier-1 U.S. carriers
  • Authored ISO 9001-compliant SOPs and trained field teams on structured troubleshooting methodologies

Fiber Optic Technician

Wyyerd Fiber | Casa Grande, AZ | 2022 - 2024

  • Integrated Lean engineering principles into field workflows, developing probability-ranked troubleshooting knowledge base that delivered $100K+ annual cost savings
  • Trained technicians on root-cause analysis (5 Whys, Fishbone) to consistently exceed SLAs

Integration Technician

SubCom | Newington, NH | 2016 - 2020

  • Re-engineered fiber inspection workflow reducing inspection time by 50% and doubling throughput on subsea cable projects
  • Designed training methodology minimizing technician errors and protecting multi-million dollar project timelines

M.S. Software Engineering

Western Governors University ยท AI Specialization

B.S. Computer Science

University of the People

Technical Stack

AI / ML

PyTorch NVIDIA NeMo Hugging Face FAISS llama.cpp scikit-learn

Models

Parakeet ASR TitaNet Gemma 3 DistilRoBERTa MiniLM

Backend

Python FastAPI PostgreSQL Redis WebSocket

Infrastructure

Docker Kubernetes CUDA Linux Nginx
Every role shows the same pattern: identify inefficiencies, build systematic solutions, document them for the team. As a Lean Six Sigma Green Belt, I combine structured process improvement with modern AI tooling. From re-engineering fiber workflows that doubled throughput to designing GPU semaphores running 22 ML engines on constrained hardware - I ship production-ready solutions and ensure the team never faces the same problem twice.

Download Resume

Download Resume (PDF)