Primary employment
Principal Engineer, AI & Automation
DocuSketch
About
Engineering intelligence into dependable systems.
I’m Shimmy Bezalel, an applied ML engineer and technical leader. I work where software, perception, physical constraints, and accountable delivery meet.
Throughline
A career built in layers
The progression is deliberate: reliable software first, then teams and technical direction, then systems that must interpret and act on the real world. The current chapter brings those disciplines into applied ML and agent workflows with explicit evidence and control.
Career arc
A systems throughline
Software → Leadership → Computer vision & 3D → Edge ML & robotics → Agent systems
- 01
Software
Production foundations across systems and interfaces.
- 02
Leadership
Teams, architecture, and accountable delivery.
- 03
Computer vision & 3D
Perception and geometry for real scenes and objects.
- 04
Edge ML & robotics
Models and measurements under physical constraints.
- 05
Agent systems
Explicit workflows, validation, and provenance.
Current
Current work
Part-time
Software & AI Lead
Robee Contech LTD
Part-time
Co-Founder
Forge Flows
How I work
Evidence, constraints, control
Evaluate before scaling
Make quality, failure modes, and evidence visible before adding reach.
Design for real constraints
Build for the devices, teams, data, and operating conditions that exist.
Keep systems inspectable
Use explicit contracts, observable runs, and meaningful human control.
Research
Learning through inspectable work
DeepSOM is a public academic project on generative manipulation for 3D texture synthesis.
View the DeepSOM case studyCurrent focus
Making system behavior inspectable
I’m focused on applied ML systems whose behavior can be inspected under deployment constraints: evaluation, failure analysis, edge behavior, and human-controlled agent workflows.
Choose a contact path