Applied ML · Computer vision · Agent systems

I build intelligent systems that hold up in the real world.

I'm Shimmy Bezalel, an applied ML engineer and technical leader working across software, computer vision, 3D, edge AI, robotics, and agentic systems—from research and prototypes to deployed products and reusable workflows.

Public record

Public evidence

  1. Debate Club public archive showing matches, judges, and an inspectable transcript

    Open source · experimental

    Debate Club

    An open match protocol for private debate agents, public judge panels, and inspectable reasoning records.

    View case study
  2. Rendered DeepSOM spatial texture sample

    Academic · complete

    DeepSOM

    An academic project exploring proxy-guided manipulation of detailed 3D forms from a single paired example.

    View case study
  3. Indoor object detection validation image with ground-truth and prediction boxes

    Technical exercise · complete

    Indoor Object Detection

    A reproducible YOLO11 baseline with sequence-aware data splitting, evaluation, and visible failure analysis.

    View case study
  4. Separate company context · active

    Forge Flows

    A part-time company context for turning successful AI work into shared organizational capability.

    View case study

Current

Current work

Primary employment

Principal Engineer, AI & Automation

DocuSketch

– Present

  • Part-time

    Software & AI Lead

    Robee Contech LTD

    – Present

  • Part-time

    Co-Founder

    Forge Flows

    – Present

Career arc

A systems throughline

Software → Leadership → Computer vision & 3D → Edge ML & robotics → Agent systems

  1. 01

    Software

    Production foundations across systems and interfaces.

  2. 02

    Leadership

    Teams, architecture, and accountable delivery.

  3. 03

    Computer vision & 3D

    Perception and geometry for real scenes and objects.

  4. 04

    Edge ML & robotics

    Models and measurements under physical constraints.

  5. 05

    Agent systems

    Explicit workflows, validation, and provenance.

How I work

Evidence, constraints, control

Separate company context · Part-time

Forge Flows

Building the systems that turn successful AI work into shared organizational capability.

  1. Observe
  2. Distill
  3. Distribute
  4. Improve

Contact

Start with the work

For engineering leadership, research, or a technically demanding collaboration, choose the right contact path.

Get in touch