Available for Opportunities

Nathan

Samson.

CS student at UMBC building AI-powered systems at scale. Published IEEE researcher. Currently crafting intelligent tools at Grassroots Analytics.

Publications

Internships

20M+

Data Points

View My Work

Based in

Baltimore, MD

Scroll

01 — About

A bit about me

Computer Science student at UMBC focused on data engineering and AI. Currently at Grassroots Analytics building a news intelligence pipeline and donor scoring tools that help political campaigns make faster decisions.

Previously built production monitoring infrastructure at Capital One and a full-stack CRM platform at OmniSyncAI. Co-author on two papers in IoT and distributed systems, accepted to IEEE PerCom 2026 and submitted to ACM CCS 2026.

IEEE

Published

UMBC

CS Student

Core Technologies

Python Go C/C++ Erlang JavaScript Java React FastAPI Node.js TensorFlow PyTorch Docker AWS Kubernetes Lambda New Relic PostgreSQL MQTT

// What drives me

const passion = {

building: "systems that handle millions of events",

researching: "IoT privacy & distributed protocols",

solving: "problems that matter at scale"

}

Python Go C/C++ Erlang JavaScript Java React FastAPI Node.js TensorFlow PyTorch Docker AWS Kubernetes Lambda New Relic PostgreSQL MQTT Python Go C/C++ Erlang JavaScript Java React FastAPI Node.js TensorFlow PyTorch Docker AWS Kubernetes Lambda New Relic PostgreSQL MQTT

02 — Experience

Where I've worked

Jan — May 2026 | DC, USA

AI Developer Intern

Grassroots Analytics

Building a campaign news intelligence pipeline that ingests from multiple APIs, ranks articles using a 5-signal hybrid scorer, and auto-generates talking points with Claude/Gemini.
Building donor segmentation and propensity scoring tools on a 20M+ record dataset to predict giving behavior.
Architecting a conversational chatbot with LLM-powered intent routing for news search and content sync.

Gemini Claude Python NewsAPI LLMs

03 — Work

Featured projects

Formally Verified Code Generation

LLM code generator using the Claude API with Dafny formal verification, achieving 92% first-try success on LeetCode problems. Built on a containerized microservice stack.

Python FastAPI React PostgreSQL Docker Dafny

AI Commerce Agent

Multi-modal product search using CLIP embeddings and FAISS vector retrieval for semantic image and text search.

FastAPI Next.js CLIP FAISS

Academic Work

IEEE PerCom 2026 — Accepted

PSMark Benchmark

Distributed pub/sub benchmarking framework with synthetic IoT workloads. 5,400 publishers, 5,900 msg/sec, 8 nodes.

Erlang C++ MQTT DDS

Code Paper

ACM CCS 2026 — Submitted

MQTT-DAP

Privacy-preserving MQTT extension enabling GDPR compliance with purpose-based access control. Modified Eclipse Mosquitto in C, under 2% CPU overhead.

C MQTT IoT GDPR

Code Benchmark