Ilia Kulikov
Research Scientist at Meta
United States
Summary
🤩
Rockstar🎓
Top SchoolIlia Kulikov is a research scientist at Meta with 11 years of experience specializing in NLP and speech-focused machine learning. He holds a PhD-level research background from NYU and a track record of internships and research roles at Google, Facebook AI, and the Courant Institute’s CILVR lab. Ilia contributes to prominent open-source toolkits such as ParlAI and fairseq, adding PyTorch compatibility, tensorboard logging, and audio/ASR features that bridge research and production workflows. His earlier work on RWTH’s returnn framework improved training optimizers and data handling for recurrent models and even added hardware-specific support, reflecting pragmatic systems-level thinking. Comfortable across deep learning research and engineering, he’s known for turning academic ideas into robust tooling that speeds experimentation. He combines rigorous academic training with hands-on contributions to widely used dialogue and sequence-to-sequence frameworks.
11 years of coding experience
5 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at New York UniversityBachelor of Science - BS, Informatics and Computer Science, Bachelor of Science - BS, Informatics and Computer Science at Ural State Technical UniversityMaster of Science - MS, Computer Software Engineering, Master of Science - MS, Computer Software Engineering at RWTH Aachen University
Github Skills (19)
hdf10
pytorch10
python10
optimizers10
optim10
machine-learning10
optimizer10
recurrent-neural-networks10
seq2seq10
deep-learning10
tensorflow10
adam10
speech-recognition10
nlp10
fairseq10
Programming languages (4)
Github contributions (5)
facebookresearch/ParlAI
May 2018 - Oct 2019
A framework for training and evaluating AI models on a variety of openly available dialogue datasets.
Role in this project:
ML Engineer
ML EngineerContributions:109 commits, 70 PRs, 87 pushes in 1 year 5 months
Contributions summary:Ilia primarily focused on improving the `parlai` framework, a dialogue AI training and evaluation framework, with updates related to PyTorch 0.4 compatibility. This involved code changes related to agents, specifically for the seq2seq and language model agents. The user also implemented features for tensorboard logging, allowing for the tracking of metrics like perplexity and loss during training and validation.
nlpdeep-learningdatasetmachine-learningtraining
facebookresearch/fairseq
Aug 2019 - Dec 2022
Facebook AI Research Sequence-to-Sequence Toolkit written in Python.
Role in this project:ML Engineer
ML EngineerContributions:5 reviews, 11 commits, 6 PRs in 3 years 4 months
Contributions summary:Ilia contributed to the fairseq repository, which focuses on sequence-to-sequence tasks within the domain of AI. Their commits show a focus on speech-related functionality, including modifying the xm transformer model, adding support for reading audio from zipped archives, and incorporating an ASR BLEU tool. These changes suggest they are involved in audio-related machine learning tasks.
pytorchnlpsequencepythontransformer-architecture
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.