AI/ML Engineer
& Creative Artist

I'm a passionate software and AI/ML engineer who enjoys solving complex problems, turning vision into code, and expressing creativity through music and visual arts. Welcome to my digital space!

About Me

A little more about who I am

Tunisian born and raised, I’m a software and machine learning engineer passionate about creating meaningful digital experiences and solving real-world problems through technology. A lifelong learner at heart, I’m constantly drawn to new ideas and collaborations that push me forward.

Over the past few years, I’ve had the opportunity to intern and work at several companies, including Amazon, where I contributed to projects spanning software engineering and AI/ML. Currently, I’m completing my Master of Engineering in Artificial Intelligence at Duke University, where I’ve explored both applied projects and research. One of my research efforts, MIRAGE, focuses on generating imperceptible audio perturbations that protect artists’ work from copyright infringement and unauthorized data scraping, reducing the risks of AI mimicry by voice retrieval models.

Outside of engineering, I immerse myself in music — producing electronic tracks, experimenting with sound design, and DJing. I’m also the cofounder of 4n Tribe, a record label and collective dedicated to amplifying diverse voices in music and advancing opportunities in sync licensing.

Ahmed Boutar Portrait

Current Experience

A snapshot of what I've been up to

2024 – Present

Duke University – MEng AI

Teaching Assistant & Research Assistant | Durham, NC


Amazon

SDE Intern | New York, NY


PocketPrep

Data Science Intern | Durham, NC

Featured Projects

A selection of some of my software engineering and ML projects

Audivine

Audivine turns Spotify listening habits and song lyrics into unique, collage-style visuals. For artists, it converts lyrics/metadata into marketing-ready artwork; for listeners, it generates personal art from your top tracks.

Stable-Diffusion React (Vite) TypeScript Python Flask Replicate AWS Spotify API

RAG Vault

A secure RAG web app: upload PDFs/DOCX/TXT, toggle retrieval on/off, and chat with foundation models on Amazon Bedrock (Claude/Titan). Per-user namespaces, JWT auth, and session history built-in.

React (Vite) Tailwind FastAPI PostgreSQL Amazon Bedrock Pinecone S3 CloudWatch

Agentic-Chatbot (Duke)

Full-stack agentic chatbot for Duke University: routes queries across curriculum, events, and places agents; augments with Pinecone RAG; uses GPT-4o-mini via LangChain; Flask backend on EC2 with a WebSocket path via AWS Lambda.

React (Vite) Flask LangChain GPT-4o-mini Pinecone AWS EC2 & Lambda
LegalBERT visualization

LegalBERT - Bill Subject classification

Fine-tuned DistilBERT to predict U.S. legislative bill subjects from titles/metadata, injecting factual legal knowledge and evaluating with robust NLP metrics and visuals

Python PyTorch Transformers Scikit-learn Fine-tuning Google Colab W&B

My Music

Sets, demos, and released music