class AlexJuma:
def __init__(self):
self.location = "Winnipeg, MB, Canada 🇰🇪 🇨🇦"
self.roles = ["AI Engineer", "3D Vision Engineer", "Full-Stack & Desktop Developer"]
self.agents = ["LangGraph", "Gemini", "RAG + pgvector", "human-in-the-loop"]
self.vision = ["Gaussian Splatting", "TRELLIS", "COLMAP", "PyTorch + CUDA"]
self.building = "multi-agent pipelines and GPU-native 3D reconstruction"
def say_hi(self):
print("Thanks for dropping by! Let's build something amazing together.")
me = AlexJuma()
me.say_hi()| 🤖 Agentic AI systems | 🧊 3D vision on the GPU |
|---|---|
| AdForge A crew of specialised agents (director, art director, cinematographer, editor, QA supervisor) that turns a product brief into consistent video ads with Veo 3.1 and Gemini 2.5. |
trellis-blackwell Microsoft's TRELLIS image-to-3D model, made to run on Linux with NVIDIA Blackwell GPUs. |
| Maiya Multi-agent RAG that ingests code, PDFs, images, audio and spreadsheets into pgvector and answers with citations. |
Gaussian Splatting pipelines Video → COLMAP → gsplat / Nerfstudio → web viewer, containerised for Blackwell GPUs. |
| LangGraph Email Reviewer Support-triage agent that classifies intent and urgency, retrieves policy, drafts replies with Gemini and waits for human approval. |
Motion-Detector Webcam motion detection with OpenCV that logs when each event starts and ends. |
Includes private work · rebuilt every 6 hours by GitHub Actions
