Hi, I’m Assem Alhomsi.

I’m an AI engineer focused on building practical generative AI systems — from RAG pipelines and LLM orchestration to backend APIs that ship to production.

Connect

Selected work

boxMind Academy — AI-Powered Learning Platform

Built an AI-driven robotics learning platform for grades 1–12 with interactive tutoring, quizzes, and lesson-level chat assistance.

Impact: Enabled self-guided learning by replacing static content with step-by-step AI tutors and adaptive quiz guidance.

  • Designed robotics curriculum for grades 1–12 in collaboration with stakeholders and school officials
  • Built backend infrastructure using Flask and MySQL to support lessons, quizzes, and user progress
  • Implemented lesson-scoped RAG system allowing students to ask questions about course material
  • Developed interactive AI tutor that guides students step-by-step instead of giving direct answers
  • Built quiz assistant that explains mistakes and leads students to correct reasoning
PythonFlaskMySQLRAGLangChainOpenAI APIDocker

boxMind Flagship App — Production GenAI Platform

Core GenAI platform powering multiple internal and customer-facing applications at boxMind.

Impact: Improved answer quality and reliability through custom model training, automated ingestion, and systematic evaluation.

  • Fine-tuned and trained custom LLMs alongside OpenAI, LLaMA, and Mistral APIs
  • Designed and implemented scalable RAG architecture using Milvus vector database
  • Automated data ingestion and embedding pipelines using Apache Airflow
  • Added RAGAS-based evaluation to compare models, prompts, and hyperparameters
  • Deployed full application stack to AWS with containerized services
PythonLangChainMilvusAirflowRAGASAWSDockerLinux

Financial Intelligence POC — NL → SQL for SWIFT Transactions

Proof-of-concept system enabling analysts to query SWIFT transaction data using natural language.

Impact: Reduced manual querying complexity by allowing non-technical users to analyze financial data via chat.

  • Built LangGraph-based agent system to translate natural language into SQL queries
  • Performed prompt engineering to improve query correctness and reasoning steps
  • Fine-tuned custom model for domain-specific financial terminology and schema understanding
  • Integrated AI chatbot for transaction analysis and follow-up questions
PythonLangGraphLangChainSQLLLM Fine-Tuning

Applied AI POCs — Healthcare, Legal, and Customer Service

Developed multiple GenAI proof-of-concepts across regulated and real-time domains.

Impact: Demonstrated feasibility of AI assistants in healthcare, legal, and customer service environments.

  • Built RAG system for hospital internal rules and regulations
  • Developed legal chatbot for document and policy querying
  • Implemented TTS/STT pipelines for stock brokers and fast-food ordering systems
  • Handled containerization, Linux setup, and deployment for rapid prototyping
  • Worked across backend, ML, and infrastructure roles in a startup environment
PythonRAGTTS/STTDockerLinuxOpenAI API
What I Build.

LLM & RAG Systems

Production RAG pipelines, vector search, and agentic workflows for real-world AI apps.

Milvus · LangChain · LangGraph · RAGAS · LangSmith · A100

Backend & APIs

Scalable backend services powering AI features, integrations, and internal tools.

Flask · REST APIs · SQL · Docker · Linux

ML Infrastructure

Deploying and operating GPU-backed model serving with reliable CI/CD and monitoring.

vLLM · Hugging Face · CI/CD · GitHub Actions · Airflow

Automation & Data Pipelines

Ingestion, embeddings, evaluation, and workflow automation to ship faster with quality.

Embeddings · preprocessing · evaluation · workflow automation

About

Overview

I’m an AI engineer who builds production-ready GenAI systems end-to-end — from ingestion and RAG pipelines to agentic workflows and the backend services that ship them. At boxMind.ai, I deployed LLMs on NVIDIA A100 infrastructure, built inference APIs, and improved retrieval quality through evaluation-driven iteration. I enjoy owning the full stack of an AI product: embeddings, retrieval, orchestration, and deployment. Based in Dearborn Heights, MI.

Context

What I focus on

Shipping production GenAI systems — not demos.

Where I add value

RAG quality, evaluation loops, and agent orchestration.

Currently

Dearborn Heights, MI · open to AI/ML roles

Let’s work together.

Portfolio by Assem Alhomsi © 2026