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

New feature

Chat with Assem

Ask an AI version of me about my projects, skills, background, and experience.

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