I'm an AI Engineer focused on Computer Vision, real-time intelligent systems, and physical-world AI.

My work combines visual perception, multi-camera geometry, deep learning, 3D reconstruction, tracking, and intelligent automation to build systems that understand and interact with real environments — not just process data on a screen.

I enjoy working at the intersection of research and engineering: understanding problems from first principles, prototyping solutions, and turning them into working end-to-end systems.

I'm also the Founder of Agentix Labs AI, where I build applied AI systems, intelligent agents, knowledge systems, and automation solutions.

Research interest: building AI systems that can observe the physical world, construct a coherent representation of it, reason over that representation, and turn perception into useful action.

A research-oriented multi-camera computer vision system that reconstructs real human motion in 3D from synchronized camera views.

The system was developed and evaluated using a table-tennis environment as a proof of concept, while the underlying architecture is designed as a domain-agnostic multi-view perception pipeline rather than a sports-specific tracking system.

Multi-Camera Capture
        │
        ▼
Camera Calibration
        │
        ▼
YOLO Detection
        │
        ▼
DeepSORT Tracking
        │
        ▼
AlphaPose Estimation
        │
        ▼
DLT Triangulation
        │
        ▼
3D Reconstruction
        │
        ▼
Unity Digital Twin

Engineering highlights

• Built a custom ~24K-image dataset
• Developed a progressive multi-stage training pipeline
• Combined detection, tracking, pose estimation, geometry, and 3D visualization
• Fused observations from independent cameras into one reconstructed physical scene
• Designed the core geometry pipeline to transfer beyond the original table-tennis environment

Core technologies

Python · OpenCV · YOLO · DeepSORT · AlphaPose · DLT · Multi-View Geometry · Unity 3D

Currently exploring an AI assistant capable of understanding software interfaces and providing real-time visual guidance directly on screen.

Instead of only answering with text, the system is intended to understand what the user is seeing and provide grounded visual assistance through contextual overlays.

Voice Command
      │
      ▼
Speech Understanding
      │
      ▼
Screen / UI Perception
      │
      ▼
Visual Reasoning
      │
      ▼
Grounded UI Element Detection
      │
      ▼
Interactive Overlay Guidance

Research challenges

UI Grounding · Vision-Language Models · Coordinate Mapping · Hallucination Reduction · Real-Time Overlay Rendering

Through Agentix Labs AI, I design and build AI agents, conversational systems, RAG pipelines, and intelligent automation using Python-first architectures.

The focus is on developing reliable AI systems that can reason, retrieve knowledge, use tools, communicate across voice and messaging channels, and execute multi-step workflows.

• Conversational & Calling AI Agents
• AI Receptionists & Booking Systems
• RAG & Knowledge Systems
• Tool-Using AI Agents
• Multi-Agent & Stateful Workflows
• Voice AI & Real-Time Conversational Systems
• WhatsApp & Business Communication Agents
• Custom AI Automation & Integrations

Python · LangGraph · LangChain · LLMs · Tool Calling · Structured Outputs · Stateful Agents

PostgreSQL · Pinecone · ChromaDB · Cohere · Vector Search · Embeddings · Semantic Retrieval

Pipecat · Asterisk · STT · TTS · Real-Time Voice Pipelines

User / Business Channel
         │
         ▼
 Voice · WhatsApp · Web
         │
         ▼
┌─────────────────────────┐
│     AI Agent Layer      │
│ Python · LangGraph · LLM│
└────────────┬────────────┘
             │
     ┌───────┼────────┐
     ▼       ▼        ▼
   Tools    RAG   State / Memory
     │       │        │
     ▼       ▼        ▼
   APIs   Vector DB  PostgreSQL
             │
             ▼
      Business Actions

AI · Computer Vision · Machine Learning

3D · Spatial Computing · Simulation

AI Systems · Backend · Automation

Engineering · DevOps

                         PHYSICAL WORLD
                               │
                               ▼
                     ┌───────────────────┐
                     │ Visual Perception │
                     └─────────┬─────────┘
                               │
                    ┌──────────┴──────────┐
                    ▼                     ▼
           Multi-Camera Vision        Spatial / 3D AI
                    │                     │
                    └──────────┬──────────┘
                               ▼
                     Scene Understanding
                               │
                               ▼
                    Intelligent Reasoning
                               │
                               ▼
                 ┌─────────────┴─────────────┐
                 ▼                           ▼
           Digital Twins                 Robotics
                 │                           │
                 └─────────────┬─────────────┘
                               ▼
                     PHYSICAL INTELLIGENCE

My long-term research interests include multi-camera intelligence, geometry-based vision, 3D scene understanding, digital twins, real-time perception, and intelligent systems capable of understanding dynamic physical environments.

🌐 Connect

Interested in Computer Vision, AI Systems, 3D Intelligence, or building something ambitious?

Building Intelligent Systems Beyond the Screen.

Computer Vision • 3D Intelligence • Physical AI • Intelligent Systems