ibraheem amin

B.S.E. Computer Science · princeton '28 · lowell, ma

Computer Science at Princeton University. I build systems at the intersection of AI tooling, web infrastructure, and developer experience.

Right now I am a software engineer intern at Knot and co-president of TigerApps, Princeton's student-run app ecosystem, while building Harness, an AI agent platform for creating, managing, and deploying custom tool configurations. Before that I was the founding software engineer at EVAL, a high school e-sports recruiting platform that connects student gamers with college opportunities.

Before Princeton, I came through the Greater Lowell Technical High School IT program, where I earned 36 industry certifications across networking, security, software development, and cloud. I have been shipping software since 2019, when I built ZBot, a Discord guild management system that served 300+ users. My 3D wildfire visualization work with NASA MODIS satellite data was presented at an MIT symposium.

██╗ █████╗
██║██╔══██╗
██║███████║
██║██╔══██║
██║██║  ██║
╚═╝╚═╝  ╚═╝

useribraheem amin (he/him)

hostprinceton university

osb.s.e. computer science '28

locationlowell, ma

uptimeshipping since 2019

pkgs7 projects · 36 certs · 13 roles

ibraheem amin — princeton, nj

stat: resume.pdf: No such file
or directory

resume is being typeset —
mail me instead.

$ mail ibraheem

connecting to the agent stream…

└─ streaming from herdr

  started   proc
● jun 2026  knot · new york
            └─ software engineer intern
● jan 2026  tigerapps · princeton, nj
            └─ co-president & senior software engineer

13 roles · 12 workplaces · 2022 → present

knot

└─ software engineer intern · jun 2026 → now

Managing cross-team automations, an agentic HTTP service and systems for managing customer escalations, and cross-team observability.

harness

● active · spring 2026

AI agent platform for creating, managing, and deploying custom tool configurations. Features real-time chat, sandbox code execution, MCP server integration, and multi-model support with context switching in seconds.

ReactTanStack StartConvexFastAPICloudflare WorkersTurborepo

github.com/DIodide/Harness ↗harness.nz ↗

Harness — landing hero, 16:9 ("Every coding agent. One chat. Every tool.")

harness-landing-hero.png

yonder-homepage.png

yonder

○ archived · fall 2025

AI-powered travel planning platform that simplifies trip planning.

Next.jsReactViteRailway

tryyonder.com ↗

dillydally-dashboard.png

dillydally

◐ hackathon · fall 2025

Productivity self-surveillance tool that tracks focus and engagement through computer vision and LLM insights.

ReactConvexExpress.jsTensorFlow

dillydallee.tech ↗DillyDally ↗DillyDallyMCP ↗demo ↗

mindbridge-title.png

mindbridge

◐ hackathon · fall 2024

Roadmap generation tool creating personalized technical and educational guides for any topic.

Next.jsOpenAI APITailwind CSSMUI

mindbridge.courses ↗MindBridge-front ↗

blockwarriors-ai.webp

blockwarriors ai

● active · 2024-2025

Java plugin system handling real-time game state management, player coordination, and integration with the web application via WebSockets for live tournament broadcasting and team management.

JavaSocket.ioNext.jsExpress.js

blockwarriors.ai ↗blockwarriors-cmd ↗

wildfiresglobe1.png

3d wildfires

○ archived · summer 2023

3D wildfire animation system using Maya and MEL scripting to visualize NASA MODIS satellite data.

Autodesk MayaPythonMEL Scripting

repo: lost

territory-taken-1.png

zbot

○ archived · 2019-2022

Premier Discord guild management system integrating with Wynncraft APIs.

PythonDiscord.pyAirflowAiohttp

repo: unpublished

* 2019  zbot — Premier Discord guild management system integrati…
* 2023  3d wildfires — 3D wildfire animation system using Maya and MEL s…
* 2024  blockwarriors ai — Java plugin system handling real-time game state…
* 2024  mindbridge — Roadmap generation tool creating personalized tec…
* 2025  dillydally — Productivity self-surveillance tool that tracks f…
* 2025  yonder — AI-powered travel planning platform that simplifi…
* 2026  harness — AI agent platform for creating, managing, and dep… (HEAD -> main)

certiport (a pearson vue business)

ciw (certified internet web professional)

cisco

microsoft

google (via coursera)

unity technologies

project management institute

testout

comptia

careersafe

Certificate:
    Data:
        Version: 3 (0x2)
        Signature Algorithm: glths-it-diploma
        Issuer: O=Certiport (A Pearson VUE business)
        Subject: CN=Ibraheem Amin
        Validity
            Not Before: Spring 2024
            Not After : does not expire
        Category: Product development & management
    Authority Information Access:
        Verify — credly.com ↗

ESB: Entrepreneurship & Small Business certificate scan

▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 15 software development & programming
▇▇▇▇▇▇▇▇▇        9 network security & infrastructure
▇▇▇▇▇            5 cloud computing & ai
▇▇▇▇             4 product development & management
▇▇               2 it fundamentals
▇                1 workplace safety

F24 ▸COS 226 MAT 201 PHY 103 FRS 149

S25 ▸COS 217 MAT 217 ORF 245 PHY 104 WRI 144

F25 ▸COS 324 COS 418 ORF 309 ANT 211

S26 ▸COS 240 COS 333 COS 484 ECO 312 HIS 387 / ENG 389 / CDH 387

~/coursework
├── cos/
│   ├── COS 226  algorithms and data structures
│   ├── COS 217  intro to programming systems
│   ├── COS 324  intro to machine learning
│   ├── COS 418  distributed systems
│   ├── COS 240  reasoning about computation
│   ├── COS 333  advanced programming techniques
│   └── COS 484  natural language processing
├── mat/
│   ├── MAT 201  multivariable calculus
│   ├── MAT 217  honors linear algebra
│   ├── ORF 245  fundamentals of statistics
│   └── ORF 309  probability and stochastic systems
├── phy/
│   ├── PHY 103  general physics 1
│   └── PHY 104  general physics 2
├── eco/
│   └── ECO 312  econometrics
└── hum/
    ├── FRS 149  ethics in finance
    ├── WRI 144  priceless
    ├── ANT 211  surveillance, technoscience, and society
    └── HIS 387  data and culture

5 directories, 18 courses

COS 418Raft Consensus & Fault-Tolerant Key–Value Store — Go

Distributed Systems · Fall 2025

Implemented the Raft consensus algorithm from the ground up in Go leader election, log replication, persistence, and snapshotting. …

▸Correct handling of the hardest distributed-systems edge cases: split votes, network partitions, stale leaders, and duplicate client requests under at-least-once RPC semantics.

▸Snapshot-based log compaction to keep memory bounded over long-running logs.

distributed consensusreplicated state machinesRPCfailure recovery

COS 333Harness — Control Plane for AI Agent Harnesses

Advanced Programming Techniques · Spring 2026 · team: Ibraheem Amin (lead), Abu Ahmed, Cole Ramer, John Wu, Richard Wang

Led a five-person team building Harness, a multi-tenant platform for creating, managing, and hot-swapping the *harness* — the full set of MCP servers, tools, and skills — behind any coding agent, without restarting the agent loop. …

▸Full-stack product: React / TanStack Start front end, Convex reactive backend, FastAPI services, and Cloudflare Workers, organized as a Turborepo monorepo.

▸Bring-your-own-agent architecture (Claude Code, Codex CLI, Cursor) built on the Agent Client Protocol, with credentials brokered server-side and agents executed in isolated, approvals-first sandboxes.

product design & team leadershipTypeScript full-stackreal-time backendsMCP/agent tooling

Harness ↗harness.nz ↗

COS 484Efficient Interactive Debugging with Learned Sequential Policies (ChatDBG RL Benchmark)

Natural Language Processing · Spring 2026 · team: Ibraheem Amin, Anika Mehrotra, Shreyas Garimella

Designed a research project extending ChatDBG (an LLM-augmented GDB/LLDB debugging assistant) that asks whether expensive, prompt-only frontier-model reasoning can be replaced by cheaper local models trained with reinforcement learning to drive a debugger. …

▸Built a benchmark comparing three control regimes: prompt-only large-model (GPT-4 / Nemotron zero-shot), an RL-trained autoregressive transformer policy, and a Mamba-based streaming-state policy that maintains persistent hidden state across debugger steps.

▸Planned an ablation across zero-shot → SFT → RLVR → RLHF (three-stage pipeline on the Tinker platform), measuring fix rate, actions per session, wall-clock latency, cost per diagnosis, and peak memory.

LLM agents & tool usereinforcement learning (RLVR/RLHF)NLP evaluation designdebugging systems

HIS 387The Orange and Black Box — Data, Recommendation, and Pastoral Power

Data and Culture · Spring 2026

Built an interactive web essay / podcast experience critically examining the rise of data-driven course selection at Princeton (Tiger Junction, Princeton Courses, and an emerging LLM recommender).

▸Pairs a produced interview transcript (student, technical, and critical perspectives) with humanities analysis, bridging STS theory and working software.

▸Demonstrates range: applying surveillance-studies and media-theory frameworks to the same agent tooling explored technically in COS 333 and COS 484.

critical technology studiestechnical writingTypeScript web appsinterdisciplinary synthesis

orange-black-box.vercel.app ↗

ECO 312ECO 312 Study Tool

Econometrics: A Mathematical Approach · Spring 2026

Built a React study/review web app for mathematical econometrics — an interactive tool for drilling estimators, distributions, and proofs — turning exam prep into a reusable, shareable piece of software.

Reactapplied statistics/econometricsself-directed tooling

eco312-review ↗

COS 226Fraud Detection

Algorithms and Data Structures · Fall 2024

description pending