Hi, I’m Peter.

Computer Science student working across software systems, cloud infrastructure, distributed systems, and AI research.

“The best way to predict the future is to invent it.”

Alan Kay

View Projects Research

Based in the NY/NJ Metro AreaOpen to internships and full-time software engineering opportunities

Portrait of Peter Ukor in a dark suit and tie

Computer Science at NJIT

Requests moving through a typical backend: cache hits return early, writes go through the queue.

A little about me.

I’m Peter, a Computer Science student minoring in Artificial Intelligence. I’m especially interested in backend engineering, distributed systems, cloud infrastructure, and AI, and I like learning by building things from the ground up.

Alongside my software projects, I’m involved in two research efforts, including working as a research assistant on RELINK. I enjoy tackling problems I haven’t seen before and understanding how systems work beneath the surface.

APIs, auth boundaries, and data models that hold up under real users.

Getting several machines to agree, even when some of them fail.

Containers, deployment pipelines, and the plumbing under every app.

Agents that reason over real evidence instead of inventing it.

Influence on social graphs and the language teens use online.

Tap a topic for the short version.

Studying: B.S. Computer Science, minor in AI, at NJIT
Currently building: Persistence and a client API for kvraft

Selected Work

Four builds, each with a live model of how it works. Poke at them: the diagrams respond.

SocraSage

An AI study companion that lets students upload their own course material, read it in the browser, and use AI to explain passages, answer questions, generate summaries, create flashcards, and build quizzes from the material they are studying.

Solo developer responsible for the architecture, backend, cloud infrastructure, AI integration, and frontend. Built around a horizontally scaled Go API, event-driven file processing, retrieval-augmented AI, and server-enforced usage controls.

Go
React
AWS
DynamoDB
S3
SQS
Lambda
Cognito
OpenAI

Live app

Hover or tap a node.

Any of the three stateless Go servers can take a request. AI answers stream back to the browser.

Guardian

An AI-powered pre-merge risk analyzer that examines dependency relationships, git history, and the potential blast radius of code changes.

Solo build for IBM TechXchange 2026. The AI explains the evidence; it never computes the score.

Python
IBM watsonx.ai
SQLite
NetworkX
Git
AI Agents

Code on GitHub

Hover a file to trace it. Click to see its risk.

src/store/files.py

High risk

Risk: 7.9
Affected: 7
Dependencies: 1

Blast radius: scanner.py, tools.py, scorer.py, git_miner.py, passport.py, cli.py, agent.py

kvraft

A distributed key-value store built from scratch using the Raft consensus algorithm.

Solo, in Go's standard library. Five real processes elect leaders and replicate a log over HTTP.

Go
Raft
RPC
Distributed Systems

Election, heartbeats, and log replication work. Persistence and the client API are next.

Code on GitHub

Term 1 · 5/5 aliveHealthy

Term 1: A is leader, sending heartbeats.

ATS for Candidates

A full-stack applicant tracking system built with a cross-functional team, with a strong focus on backend architecture, authorization, testing, and database design.

Backend and authentication on a six-person team. Deployed, with 556 tests gating every pull request.

Next.js
React
TypeScript
PostgreSQL
Prisma
GitHub Actions

Live app Code on GitHub

Follow a request from candidate to deploy. Hover or tap a stage.

Research

Two projects on how information and language move through online communities: one on graphs, one on words.

RELINK

Research on deep reinforcement learning for edge-level influence maximization in closed social networks.

Instead of choosing which people to start with, the agent learns which connections to activate so influence travels furthest.

Role: Research Assistant
Supervision: Under the lead professor

activated selected edge10 of 24 reached

Hover or tap a person to trace how influence reached them.

Teen Mental Health Language Detection

NLP research focused on adolescent digital language, dataset construction, annotation, and model evaluation.

examples annotated and preprocessed: 20,000; examples annotated and preprocessed
severity scale in the taxonomy: 0–4; severity scale in the taxonomy
Krippendorff agreement tracked: α ≥ 0.70; Krippendorff agreement tracked

Role: Research Assistant
Supervision: Under a PhD student

From raw posts to a measured model. Hover or tap a stage.

Experience

Education

New Jersey Institute of Technology

B.S. Computer Science

Minor in Artificial Intelligence

Research Assistant, RELINK
Current
Deep reinforcement learning for edge-level influence maximization in closed social networks.
Research Assistant, Teen Mental Health Language Detection
2026 – present
Annotating and preprocessing a 20,000-example NLP dataset across four platforms.
Manufacturing Operator, Integra LifeSciences
2023 – 2024
Automated data-entry validation that cut record-keeping errors by more than 20% across a 10-person team, under FDA standards.
Member of ColorStack and NSBE at NJIT, and a Year Up United participant.

Skills

Hover or tap a skill to see where I used it.

Languages

Backend & Web

Cloud, Systems & AI

Want to build something interesting?

I’m based in the NY/NJ Metro Area and currently open to internship and full-time software engineering opportunities.

peter@peterukor.com