Software Engineering Intern @ Princeton · CS @ Rutgers '28

Backend and cloud,
built to last.

I build the services behind the product: Linux daemons, secure APIs, data pipelines, and ML models in production. Scroll to travel through the work.

developer.ts

const abubaker = {
  role:  'SWE Intern @ Princeton',
  study: 'CS @ Rutgers, May 2028',
  focus: ['backend', 'cloud',
          'data', 'ML services'],
  stack: ['Python', 'Linux', 'SQL',
          'Node.js', 'Docker'],
  status: 'open to opportunities',
}  // this page is one scene, keep scrolling

01 · Princeton University · Jul 2026 to present

Zero-touch file access for 200+ fab researchers

As a Software Engineering Intern I am building an event-driven Python daemon (systemd, Linux) that provisions per-researcher file access on demand: temporary POSIX ACLs and bind-mounts applied at tool login and revoked at logout, replacing a manual workflow.

A companion open-source plugin and HTTP API translate tool sessions into provisioning events over mutual TLS, backed by a WAL-mode SQLite session store. I hardened the service with interface binding, HTTPS, and hashed API keys, then wrote its requirements and architecture documentation for the university IT team.

Python
Linux / systemd
mTLS
SQLite
HTTP API

02 · Data pipelines & ML services

Models and pipelines that run in production

Shipment Weight Prediction Service. Trained on a full year of warehouse data to predict shipment weights at 95% accuracy for an operation moving 10,000+ shipments a day. Served through a FastAPI REST service and packaged as a reusable Python library, cutting roughly 28 hours of manual weighing a day.

Mercer Council, Software Developer (2024 to 2026). Designed a relational schema and Python ingestion pipeline that replaced multi-day manual reporting, ran a cloud OCR pipeline over 7 years and 140+ quarterly documents into a queryable dataset, and shipped a live dashboard for the executive team.

RAISE '26 national AI research finalist, 1 of 8 teams from 188: an NLP pipeline over 10,500+ articles using local LLMs, embeddings, clustering, and sentiment analysis.

FastAPI
scikit-learn
Pandas
SQL
OCR

03 · Full-stack platforms

Systems that carry real traffic

ClubSphere is a club operations platform now used by 600+ students across 18 clubs: member management, event tracking, finances, document storage, and real-time messaging over WebSockets, on a React and Node.js stack that consolidated fragmented campus workflows into one system.

CSAI.club scaled to 227 monthly users, with an automated board that ingests and prunes listings weekly to hold 200+ live New Jersey internship postings.

I also co-founded and chair the National Community of Practice (cophub.org), a network of 19+ colleges and universities across the United States.

Node.js
React
WebSocket
Python

clubsphere.csai.club → csai.club → cophub.org →

04 · Princeton Materials Institute · Dec 2024 to Aug 2025

A training platform with three LLM agents behind it

I built the chip-packaging training platform used to train 150+ technicians, and coordinated a 6-person team. My work centred on the service layer: three LLM agents (an Instructor, an Evaluator, and a Safety Monitor) integrated through the OpenAI API to guide and assess trainees in real time, backed by a system storing per-user performance analytics.

The simulation itself reproduced the semiconductor packaging process via photogrammetry, giving technicians hands-on practice without physical access to the tools.

OpenAI API
Analytics
Unity / C#
Photogrammetry

Published, J. Advanced Technological Education →

05 · Side work

Making invisible physics visible

My Physics Simulation Suite is a collection of 30+ interactive electromagnetic simulations covering field lines, current flow, and the Lorentz force spiral you see here, built as teaching tools for physics students.

At Rutgers I serve as an Adobe Student Ambassador, running campus creative-skill and student engagement initiatives.