MLIR Maintainers

This file is a list of the maintainers for MLIR.

The following people are the active maintainers for the project. For the sake of simplicity, responsibility areas are subdivided into broad categories, which are further subdivided into individual components, such as dialects. Please reach out to them for code reviews, questions about their area of expertise, or other assistance.

Core

Core components of MLIR, including core IR, analyses and rewriters, fundamental dialects, build system and language bindings.

Alex Zinenko
ftynse@gmail.com (email), @ftynse (GitHub), ftynse (Discourse)
Jacques Pienaar
jpienaar@google.com (email), @jpienaar (GitHub), jpienaar (Discourse)
Mehdi Amini
joker.eph@gmail.com (email), @joker-eph (GitHub), mehdi_amini (Discourse)

Code

Standalone subcategories

Core tooling (ODS, DRR, PDLL, LSP) (core)
CMake (christopherbate)
Dialect Conversion (matthias-springer, zero9178)
Python Bindings (makslevental, rolfmorel)

Dialects

Code Structure Dialects

Builtin Dialect (core)
‘func’ Dialect (core)
‘scf’ Dialect (core)
‘cf’ Dialect (core)
‘index’ Dialect (core)
‘ptr’ Dialect (fabianmcg)

Basic Compute Dialects

‘arith’ Dialect (core)
‘math’ Dialect (core)
Rewrite System Dialects (core)
Transform Dialect (martin-luecke, ftynse, rolfmorel)
‘pdl_interp’ Dialect (jpienaar)
‘pdl’ Dialect (jpienaar)

Accessory Dialects

‘affine’ Dialect (ftynse)
‘dlti’ Dialect (rolfmorel)
‘irdl’ Dialect (math-fehr, moxinilian)
‘shape’ Dialect (jpienaar)
‘smt’ Dialect (fabianschuiki, maerhart)
‘ub’ Dialect (Hardcode84)
‘complex’ Dialect (core)
‘async’ Dialect (unmaintained)

Egress

MLIR components pertaining to egress flows from MLIR, in particular to LLVM IR.

Matthias Springer
me@m-sp.org (email), @matthias-springer (GitHub), matthias-springer (Discourse)
Andrzej Warzynski
andrzej.warzynski@arm.com (email), @banach-space (GitHub), banach-space (Discourse)
Tobias Gysi
tobias.gysi@nextsilicon.com (email), @gysit (GitHub), gysit (Discourse)

Dialects

The egress maintainer refers to the people working in the Egress category, with the point-of-contact being the maintainers above. Named maintainers, if available, should be contacted first, as they're more active in those areas.

Lowering Dialects

‘llvm’ Dialect (egress)
‘SPIR-V’ Dialect (@kuhar, @antiagainst)
‘emitc’ Dialect (@aniragil, @marbre)

GPU Dialects

‘gpu’ Dialect (@fabianmcg)
‘amdgpu’ Dialect (@krzysz00)
‘rocdl’ Dialect (@krzysz00)
‘nvgpu’ Dialect (@grypp)
‘nvvm’ Dialect (@grypp)
‘xegpu’ Dialect (@chencha3, @Jianhui-Li)
‘xevm’ Dialect (@silee2)

CPU Dialects

‘arm_neon’ Dialect (@banach-space)
‘arm_sve’ Dialect (@banach-space)
‘ArmSME’ Dialect (@banach-space)
‘amx’ Dialect (@adam-smnk)
‘x86vector’ Dialect (@adam-smnk)
‘vcix’ Dialect (@mshockwave)

Paradigm Dialects

‘omp’ Dialect (@tblah, @skatrak)
‘acc’ Dialect (@clementval, @razvanlupusoru)
‘mpi’ Dialect (@fschlimb)
‘shard’ Dialect (@fschlimb)

Tensor Compiler

MLIR components specific to construction of compilers for tensor algebra, in particular for machine learning compilers.

Renato Golin
rengolin@gmail.com (email), @rengolin (GitHub), rengolin (Discourse)
Jacques Pienaar
jpienaar@google.com (email), @jpienaar (GitHub), jpienaar (Discourse)
Andrzej Warzynski
andrzej.warzynski@arm.com (email), @banach-space (GitHub), banach-space (Discourse)

Dialects

The tensor maintainer refers to the people working in the tensor compiler category, with the point-of-contact being the maintainers above. These are key MLIR dialects that will never become unmaintained. Named maintainers, if available, should be contacted first, as they're more active in those areas.

Linear Algebra Dialects

‘linalg’ Dialect (tensor)
Tensor Operator Set Architecture (TOSA) Dialect (@sjarus)

Type Dialects

‘tensor’ Dialect (tensor)
‘memref’ Dialect (tensor)
‘vector’ Dialect (tensor + @dcaballe, @Groverkss)
‘sparse_tensor’ Dialect (@aartbik, @matthias-springer)

Accessory Dialects

‘bufferization’ Dialect (tensor, @matthias-springer)
‘ml_program’ Dialect (@jpienaar)
‘quant’ Dialect (unmaintained)