Skip to content
View kingardor's full-sized avatar
🚀
Diving Deeper Into Deep Learning
🚀
Diving Deeper Into Deep Learning

Organizations

@GLUG-REVA @NullPointersInc

Block or report kingardor

Report abuse

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

Report abuse
kingardor/README.md

Akash James — Banner

Portfolio Email GitHub LinkedIn X Instagram Medium

AboutFeaturedStacksStatsPostsContact


🚀 About

Generative AI ArchitectZ by HP Global Data Science AmbassadorNVIDIA Jetson AI Labex-Researcher, UC Berkeley

  • Founding AI Architect (Stealth, Mar 2025—Present): building a video-first data lake with agentic multimodal RAG & graph-native retrieval
  • Prev: Lead AI Architect @ SparkCognition
  • Focus: Qdrant (hybrid sparse+dense), GraphRAG, CrewAI orchestration, FastAPI + SSE, DeepStream/TensorRT, Docker
  • Human bit: open to collabs & advisory • when not coding, I beatbox 🎤

I like systems that move money and move needles.


🧨 Featured Builds

Hermes — Wildfire Detection
Jetson + DeepStream + YOLO; on-device CV for rapid wildfire cues.
Stars  Forks
Activity-Recognition-TensorRT
3D ResNet video classification, TensorRT-accelerated pipelines.
Stars  Forks
YOLOv4-OpenCV-CUDA-DNN
Run YOLOv4 via OpenCV’s CUDA DNN — practical, fast, portable.
Stars  Forks
Vector / RAG Utilities
Qdrant hybrid search, no-preset-k clustering, GraphRAG scaffolds.
Explore

🧩 Stacks I Ship With


📈 Stats & Streaks

GitHub Streak

Profile Views


📝 Latest Posts

➡️ more posts…


🤝 Contact

akashjamesofficial@gmail.comPortfolioGitHubInstagramLinkedInX


Badges via shields.io. “Twitter” links to X. Featured work points to public repos/demos. If you’re here for **RAG, GraphRAG, DeepStream, or CrewAI** collabs — send a 3-line brief + timeline to my email.

Pinned Loading

  1. vector-advanced-ai vector-advanced-ai Public

    Vector with Advanced AI compatibilities

    Python 8 2

  2. Hermes-Deepstream Hermes-Deepstream Public

    Hermes is a Wildfire detection system that utilizes Computer Vision and is accelerated using NVIDIA Deepstream

    C++ 66 25

  3. Activity-Recognition-TensorRT Activity-Recognition-TensorRT Public

    3D ResNet Video Classification accelerated by TensorRT

    Python 48 12

  4. YOLOv4-OpenCV-CUDA-DNN YOLOv4-OpenCV-CUDA-DNN Public

    Run YOLOv4 directly with OpenCV using the CUDA enabled DNN module.

    Python 37 12

  5. Centerface-Deepstream Centerface-Deepstream Public

    Centerface ONNX accelerated with Deepstream 5.1

    C++ 10 1

  6. ShredFPSOpponents ShredFPSOpponents Public

    Pipeline Game Input to YOLOv3 Object Detection to detect opponents.

    Python 9 2