Matthew Fitzgerald

Software Engineer | Full Stack, Backend, Infra | ML Systems

I build APIs, data pipelines, and ML systems, and I run the infrastructure they ship on.

Education

Florida Tech - B.S. in Computer Science, 2021-2024

Work Experience

Junior Full Stack Engineer at EarthCam - February 2026 - May 2026

  • Re-architected the data-fetching layer of a core customer-facing component around asynchronous API calls, cutting load time from 15,000+ ms to 2,000-3,000 ms, a 3-5x improvement
  • Integrated data-querying APIs across the application, improving data resolution and cutting redundant round-trips to lower latency
  • Delivered new pages and feature work across the stack in TypeScript and Node.js

Software Engineer (Platform & ML Infrastructure) at Cognitive Network Solutions - February 2025 - November 2025

  • Developed end-to-end machine learning pipelines in Python for telecom network optimization, spanning feature engineering (pandas, Polars), model training (PyTorch, graph neural networks, reinforcement learning), MLflow experiment tracking, and GPU-accelerated inference
  • Built streaming data pipelines with Kafka for real-time network telemetry, including dead-letter-queue replay for fault-tolerant ingestion
  • Wrote Python automation for data quality checks and security scans (exposed keys, vulnerabilities, auth misconfigurations), gating merges through GitLab CI/CD with branch-based workflows and peer code review
  • Developed and secured SQL and graph databases (PostgreSQL, Neo4j) with role-based access controls, integrated into Python microservices via SQLAlchemy, psycopg2, and FastAPI
  • Participated in the NVIDIA 6G Developer Program, building GPU-accelerated training and inference workflows (CUDA, PyTorch, TensorFlow) with auto-scaling batch jobs to optimize compute costs
  • Designed and deployed the multi-cloud infrastructure these systems ran on: Terraform across GCP and Azure, Kubernetes with Helm, and autoscaling GPU node pools for training and inference workloads

Software Engineer, Intern at Dfinitiv.io - Summer 2023, 2024

  • Engineered secure, cloud-native pipelines on AWS and GCP to automate ingestion and curation of digital media assets, reducing processing time by over 60%
  • Built and maintained asset metadata databases in PostgreSQL and MongoDB, enabling fast, reliable querying across thousands of records
  • Deployed applications and microservices using boto3, google-cloud-storage, psycopg2, and pymongo, ensuring scalability and portability
  • Built automated web scraping pipelines using Selenium and Playwright to gather and structure data from publicly available sources

About Me

I'm a software engineer who builds APIs, data pipelines, and ML systems in Python and TypeScript. At EarthCam, I re-architected the data-fetching layer of a core customer-facing component around asynchronous API calls, cutting load time from 15,000+ ms to 2,000-3,000 ms, a 3-5x improvement, and delivered new pages and features across the stack in TypeScript and Node.js.

On my own time, I build end-to-end ML systems: a fine-tuned LLM served behind a streaming FastAPI inference API with an automated evaluation gate, a ReAct-style agent with multi-turn memory and live reasoning-step streaming, and a feature engineering pipeline for financial transaction data with point-in-time correct serving and unit tests covering ingestion through serving.

At Cognitive Network Solutions, I developed end-to-end ML pipelines in Python for telecom network optimization: feature engineering with pandas and Polars, model training with PyTorch and graph neural networks, and GPU-accelerated inference tracked in MLflow. I built streaming data pipelines with Kafka for real-time network telemetry, and secured SQL and graph databases (PostgreSQL, Neo4j) integrated into Python microservices via SQLAlchemy, psycopg2, and FastAPI. I also run the infrastructure my software ships on: automating data-quality and security scans that gate every merge through GitLab CI/CD, and building GPU-accelerated training and inference workloads through the NVIDIA 6G Developer Program with auto-scaling batch jobs to manage compute costs.

Previously at Dfinitiv, I built cloud-native data pipelines on AWS and GCP that cut processing time by over 60%.

I'm looking for Software Engineer and Backend Engineer roles where I can build production software, backed by the platform experience to run it reliably.

Skills

Python (Primary)

Python FastAPI SQLAlchemy APScheduler Pandera Prefect boto3 psycopg2 pymongo Selenium Playwright

Other Languages

TypeScript/Node.js Go SQL MongoDB Java C++ Bash

ML & Data

MLX / mlx-lm Hugging Face LoRA Fine-Tuning MLflow CUDA River (Online ML) sentence-transformers ChromaDB BM25 / Hybrid Retrieval Redis GPU Node-Pool Provisioning Inference Auto-Scaling

Cloud & DevOps

Kubernetes Helm Docker Ingress Controllers Namespace Isolation Terraform GitLab CI/CD Cloud Build AWS (Lambda, S3, IAM, Secrets Manager) GCP (GKE Autopilot, Cloud Run, IAM, Secret Manager) Azure (AKS, ACR, Blob Storage) Linux Administration VPC Networking Private Clusters VPN / IAP TLS / HTTPS DNS Kong API Gateway IAM / RBAC Workload Identity Secrets Management Prometheus Structured Logging Liveness / Readiness Probes SLI/SLO Instrumentation Incident Detection Time-to-Detection Reduction Auto-Scaling On-Call-Ready Monitoring

AI-Assisted Engineering

Claude Code Agentic Workflows MCP Server Development PR-Based AI Review Loops

Projects

LLM Fine-Tuning Pipeline

Fine-tuned Mistral-7B on financial news with LoRA adapters, served as a streaming inference API with SSE token delivery and an automated ROUGE-L evaluation gate.

Libraries Used: MLX, mlx-lm, FastAPI, LoRA, Hugging Face

View Repository

LLM Agent

ReAct-style agent with tools spanning a RAG pipeline, feature store, and drift monitor, with multi-turn conversation memory and live SSE streaming of each reasoning step.

Libraries Used: FastAPI, mlx-lm, SQLAlchemy, PostgreSQL, httpx

View Repository

ML Data Pipeline

End-to-end feature engineering pipeline for financial transaction data: schema validation, windowed aggregations enforcing point-in-time correctness, a versioned offline feature store, and a FastAPI serving layer, with unit tests covering ingestion, features, store, and serving.

Libraries Used: FastAPI, Pandera, Prefect, SQLAlchemy, PostgreSQL, Redis, pandas

View Repository

ML Drift Monitor

Data drift detection service using ADWIN with an automated retraining pipeline, monitoring model inputs in production and triggering retraining on distribution shift.

Libraries Used: FastAPI, River, MLflow, APScheduler, PostgreSQL

View Repository

RAG Pipeline

Hybrid retrieval pipeline combining BM25 keyword search with dense vector retrieval, cross-encoder reranking, SSE token streaming, and per-sentence citation tracking.

Libraries Used: FastAPI, ChromaDB, BM25, sentence-transformers, mlx-lm

View Repository

LLM Guardrails

Security proxy layer in front of the LLM agent enforcing semantic injection detection, PII scrubbing, per-client rate limit tiers, replay protection, and full audit logging.

Libraries Used: FastAPI, Redis, PostgreSQL, sentence-transformers, SQLAlchemy

View Repository