Romain Liautaud
Software Engineer at nabla
Melbourne, Victoria, Australia
Summary
🤩
RockstarRomain Liautaud is a software engineer based in Melbourne with nine years of experience building robust backend and full-stack tooling, currently contributing at Nabla. He has a strong open-source track record, including work on sonos/tract where he added a CLI, profiling, visualization, and protobuf graph serialization for TensorFlow/ONNX inference, and contributions to mirage/irmin improving type systems and JSON encoding edge cases. Comfortable across languages and layers, he blends low-level data correctness (handling NaN/inf and nested option decoding) with developer ergonomics like CLI arg parsing and visual graph analysis. Colleagues can expect a pragmatic engineer who surfaces hard-to-debug performance details and ships practical tooling that improves observability and usability.
9 years of coding experience
Github Skills (24)
json10
command-line-interface10
onnx10
decoding10
ocaml10
type-system10
tensorflow10
encoding10
command-line10
json-encode10
json-encoder10
jsondecoder10
cli10
testing9
vue9
Programming languages (11)
Github contributions (5)
mirage/irmin
Feb 2020 - Jul 2020
Irmin is a distributed database that follows the same design principles as Git
Role in this project:
Back-end Developer
Back-end DeveloperContributions:47 commits, 5 PRs, 32 comments in 4 months
Contributions summary:Romain primarily contributed to the `irmin/irmin` repository by improving the type system and its associated JSON encoding. Their work included ensuring unique names for record fields, variants, and enums, and updating the pretty-printer to generate a paste-able representation of the data structures. They also fixed issues with JSON decoding, particularly around nested options, and enhanced the encoding of NaN, -inf, and +inf values. Additionally, the user added several regression tests to validate the changes.
design-principlesprinciplesocamldistributed-databasedatabase
sonos/tract
May 2018 - Jul 2018
Tiny, no-nonsense, self-contained, Tensorflow and ONNX inference
Role in this project:Full-stack Developer
Full-stack DeveloperContributions:7 commits, 10 PRs, 79 pushes in 1 month
Contributions summary:Romain implemented a command-line interface (CLI) for the project, adding features for comparing and profiling the `tfdeploy` library with TensorFlow. They introduced the `clap` crate for argument parsing and integrated logging with `simplelog`. Furthermore, the user added features for detailed profiling, including execution time analysis and visualization of node performance, contributing to the project's usability and debugging capabilities. They also implemented graph analysis and visualization functionalities using Vue, Vuetify, and Cytoscape, as well as protobuf based graph serialization.
rust-libraryno-nonsenseself-containedrustdeep-learning
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.