I'm Martin, a Machine Learning Engineer (MLE) with a focus in applied machine learning, MLOps, and data engineering. I also have experience working as a software engineer (mostly backend) through various projects and work experience.

My work ranges from AI systems & pipelines, model analysis, machine learning projects, physics simulations, game design, and full-stack development.

Currently, I am working on building an Augmented Reality application for OIS Designs that reconstructs scanned real-world environments into a 3D virtual map.

Feel free to reach me at tejada.mrt@gmail.com, and take a look around my repositories!

Computer Vision & Spatial Computing Engineer @ OIS Designs

• Built a 3D spatial mapping pipeline for an iOS AR application, converting camera input into a structured 3D reconstructed room.
• Implemented object detection and spatial placement logic enabling users to add, move, and manipulate furniture within the reconstructed 3D environment.

Backend Engineer @ Dazia Consulting Inc.

• Expanded and deployed an AI-powered learning platform from prototype to production in JavaScript with Node.js.
• Built and improved backend systems, including multi-turn conversational memory, tested and ensured application behaviors works as intended.
• Designed the application database (with Supabase) and implemented CRUD APIs for chat, flashcards, and quiz features.

AI/ML Engineer Extern @ Extern hosted by Outamation (Project: Advanced AI-Powered Document Intelligence & Data Extraction)

• Built AI-powered document processing pipelines for data extraction, classification, and semantic search using Python, with OCR systems and structured parsing of mortgage documents.
• Developed a RAG system using LlamaIndex and embedding-based search for contextual Q&A for users.
• Conducted end-to-end evaluation of system performance on real-world mortgage documents, comparing model outputs and analyzing extraction accuracy, retrieval quality, and reliability across document types.

Malicious URL Detector

• Trained a Random Forest classifier on 6,000 URLs to detect malicious links with a recall of 0.88 for malicious URLs
• Used a sentence embedding model to convert raw URL strings into numerical vectors for model input

Linear Regression Model Recreation

• Recreated the Linear Regression model (LRM) from scratch in Python and validated results with scikit-learn's LRM.
• Implemented gradient descent and Mean Squared Error in a second rendition of the model.

Gambler's Fallacy and The Regression Towards the Mean

• Built a Python simulation using NumPy and plotlib to demonstrate regression toward the mean, and clarified how it differs from the gambler's fallacy.

3D Simulation Engine (ongoing development)

• A first-person simulation engine written in C using OpenGL.
  
• Includes original camera movement, AABB collision detection, and hitscan libraries.

Context-Aware AI Chatbot

• Desktop chatbot with multi-turn memory built using both Python and OpenAI API.
  
• Supports speech-to-text translations, featuring an animated avatar and a frontend built with PyQt6.

Kaggle • Pandas • Joblib • PyMuPDF • Easy OCR • Llama Index • Hugging Face • numpy • plotlib • scikit-learn

FastAPI • Supabase • Express.js • Node.js • uvicorn

OpenGL • Git • VS Code • Eclipse IDE • Tkinter • Gradio • PyQt • Thunder Client/Postman