Skip to content
View Murci20965's full-sized avatar
🇿🇦
on the last bus between dreams and reality
🇿🇦
on the last bus between dreams and reality
2 followers · 1 following

Achievements

Achievement: Pair Extraordinairex4Achievement: Pull Sharkx2Achievement: YOLO

Achievements

Achievement: Pair Extraordinairex4Achievement: Pull Sharkx2Achievement: YOLO

Block or report Murci20965

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Murci20965/README.md
Nhlanhla Mokoena, AI Engineer: agentic AI, RAG and MLOps. Building XR learning anyone can reach. A short loop of the film from his portfolio, from a prompt to people: a night desk, a laptop screen, an agent network, glass cubes, a data centre, an XR classroom and the globe.

AI Engineer at Nudle, XR simulations. I engineer the generative pipelines behind XR simulation learning: a headless Blender agent framework that builds 3D lesson environments, multimodal text- and image-to-3D generation, text-to-video lesson animation, and the containerised GPU serving underneath.

Traditional education gates real skills behind resources and rigid methods. I'm building toward learning where anyone, anywhere, can practise real skills interactively.

02 · Live: two pipelines, live in the browser

Avatar-3D Pipeline, live demo: words to 14 animation states in React Three Fiber Orbit-3D Asset Pipeline, live demo: text or image to web-ready 3D with headless Blender

Avatar-3D Pipeline: natural language to 14 validated 3D animation states. An LLM router under strict Pydantic validation, rendered in Next.js and React Three Fiber. Code

Orbit-3D Asset Pipeline: text or image to web-ready 3D. Tripo3D generation, a vision model for context, and a Dockerised headless Blender engine that Draco-compresses each mesh. Code

03 · Open source: public from the first commit

  • Real Estate Price Predictor: XGBoost, R² 0.9037 and RMSE 0.1341 (log scale), served by FastAPI, shipped to AWS by GitHub Actions.
  • Medical Image Classifier: ResNet50 transfer learning for pneumonia detection, 82.85% accuracy and 0.96 recall.
  • Cat vs Dog Classifier: a FastAI CNN taken end to end, served by FastAPI with a Gradio UI.
  • Resume-Match AI: LLM fit scoring between a resume and a job posting.
  • Smart-Spend: a personal-finance API on FastAPI, PostgreSQL, Redis and Hugging Face models.

04 · Stack: from model to classroom

The stack, built from the ground up to the classroom. 01 Data and Libraries: Pandas, NumPy, Matplotlib, SQL, ETL pipelines. 02 ML and Deep Learning: PyTorch, Scikit-Learn, Weights and Biases, fine-tuning, inference and evaluation. 03 Agentic AI and GenAI: LangChain, LangGraph, n8n, RAG pipelines, vector databases, multi-agent workflows, Claude and OpenAI APIs. 04 Frameworks and Languages: Python, FastAPI, Next.js, React, Node.js. 05 MLOps and Delivery: Docker, Git, GitHub Actions CI/CD, end-to-end MLOps pipelines, serverless. 06 Cloud and Tools: AWS, Microsoft Azure, Hugging Face Spaces, Render, Vercel. 07 Architecture and Systems: system design, scalable architecture, microservices. 08 XR and 3D: WebXR, React Three Fiber, headless Blender, TRELLIS.
The stack as text
Layer Tools
08 XR & 3D WebXR · React Three Fiber · headless Blender · TRELLIS
07 Architecture & Systems System design · scalable architecture · microservices
06 Cloud & Tools AWS · Microsoft Azure · Hugging Face Spaces · Render · Vercel
05 MLOps & Delivery Docker · Git · GitHub Actions CI/CD · end-to-end MLOps pipelines · serverless
04 Frameworks & Languages Python · FastAPI · Next.js · React · Node.js
03 Agentic AI & GenAI LangChain · LangGraph · n8n · RAG pipelines · vector databases · multi-agent workflows · Claude & OpenAI APIs
02 ML & Deep Learning PyTorch · Scikit-Learn · Weights & Biases · fine-tuning · inference & evaluation
01 Data & Libraries Pandas · NumPy · Matplotlib · SQL · ETL pipelines

05 · Contact: let's work together. A role, a collaboration, or a question about my work. Opens the portfolio.

Portfolio · LinkedIn · Email

Set in Fraunces, Hanken Grotesk and Space Mono. The frames are from the film on my portfolio.

Pinned Loading

  1. real_estate_price_predictor real_estate_price_predictor Public

    An end-to-end MLOps project to predict house prices using XGBoost, FastAPI, Docker, and AWS.

    Jupyter Notebook 1

  2. medical_image_classifier medical_image_classifier Public

    End-to-end pneumonia detection system using PyTorch, FastAPI, and Docker. Implements a full MLOps pipeline from training and evaluation to deployment ready application with an interactive Streamlit…

    Python 1

  3. Loan_Default_Analysis. Loan_Default_Analysis. Public

    This project analyzes a bank’s customer loan data to identify key predictors of loan default

    Jupyter Notebook 1

  4. YouTube-Streamers-Analysis YouTube-Streamers-Analysis Public

    Performed a comprehensive analysis of the top YouTube streamers dataset.

    Jupyter Notebook 1

  5. Movie_recommender_system Movie_recommender_system Public

    This project implements a movie recommendation system using collaborative filtering techniques. The system analyzes user preferences and behavior to generate personalized movie recommendations, enh…

    Jupyter Notebook

  6. PowerBI-SQL-Pizza-Sales-Analysis PowerBI-SQL-Pizza-Sales-Analysis Public

    An interactive dashboard built using Power BI and SQL to analyze 2015 pizza sales data. It features key metrics such as total revenue, average order value, and top-selling pizzas, providing actiona…