DALIZHEresearch · engineering
Tehran
--:--:--
hold the button and sweep across the nodes
DALIZHE

DALIZHE

Intelligence, from model to machine
Debrief

You just moved through the systems behind intelligent machines.

The six stages of the sequence, in the order they play

  1. 01Intelligence

    Take the network

    Models begin with propagation — information moving through a learned structure.

  2. 02Intelligence

    Purge the trojans

    Language models decide what context matters before they respond.

  3. 03Robotics

    Shape the path

    Before a machine can act it needs to know where it is, and where it should go.

  4. 04Robotics

    Work the cell

    Action needs more than an instruction: planning, kinematics, feedback, control.

  5. 05Compute

    Synchronise the swarm

    Compute is defined less by its components than by how work moves between them.

  6. 06Compute

    Hold the lattice

    Performance is alignment — model, runtime, memory and hardware as one system.

At Dalizhe these are not separate demonstrations. They are connected parts of one problem: turning intelligence into machines that work reliably outside a laboratory.

Every stage you played is a simplified version of something we build.
The system

We build intelligence from model to machine.

Dalizhe is a research and engineering company working across machine learning, robotics, and computing systems. We develop the software that models reason with, the infrastructure that runs it, and the physical machines that act on its output.

Most groups take one layer of that path. We think the difficult and interesting problems sit between the layers — where a model meets a runtime, where a runtime meets hardware, and where all of it meets a machine that has to move correctly the first time.

Areas of practice

Three practices, one stack.

Stages 01–02

Intelligence

Language-model systems and the infrastructure around them — adaptation, retrieval, evaluation and inference — built as software that has to keep working, not as demonstrations.

Stages 03–04

Robotics

Robotic systems and the control stack behind them: perception, planning, inverse kinematics and trajectory generation, together with the measurement and control instrumentation they depend on.

Stages 05–06

Compute

The computing layer that makes the rest practical — networks, data-centre systems, computer and electronic hardware, and the runtimes that map workloads onto them efficiently.

The work

Experiments, prototypes and systems in development.

A selection of work across intelligence, robotics and compute. Each entry is labelled with its real status.

  • Intelligence

    Project name

    One sentence on what it does and what question it answers.

    Status
  • Robotics

    Project name

    One sentence on what it does and what question it answers.

    Status
  • Compute

    Project name

    One sentence on what it does and what question it answers.

    Status
The people

The people behind Dalizhe.

A small group of researchers and engineers working across machine learning, robotics, software and computing systems. We take on problems that need more than one discipline, rather than optimising inside a single one.

Danial Fathi
Chief Executive Officer
Mehdi Taraghijah
Chairman of the Board
Contact

Have a difficult problem in mind?

We are interested in ambitious work involving intelligent systems, robotics, and the infrastructure that makes them possible.

hello@dalizhe.ir