Kartikay Khandelwal
CTO at WaveForms AI
Palo Alto, California, United States
Summary
🤩
Rockstar🎓
Top SchoolKartikay Khandelwal is a CTO and AI-native engineering leader with seven years focused on building and scaling ML infra and multimodal systems, most recently leading WaveForms AI after senior roles driving PyTorch and AI infrastructure at Meta. He has deep hands-on experience implementing core LLM components—such as Rotary Positional Embeddings in torchtune—and significant contributions to flagship open-source toolkits like fairseq and pytext for masked language modeling and multilingual data pipelines. Comfortable bridging research and production, he has a track record of refactoring and architecting transformer encoders and efficient data samplers to make large-scale training practical. Based in Palo Alto and grounded by an MS from Stanford, he combines practitioner-level coding with strategic leadership across both startup and large-company environments. An under-the-radar strength is his knack for surfacing and fixing hard data-handling and memory issues that materially improve model training stability at scale.
7 years of coding experience
13 years of employment as a software developer
B.E. (Hons), B.E. (Hons) at Birla Institute of Technology and ScienceMS, MS at Stanford University
Github Skills (15)
transformer-models10
data-handling10
pytorch10
machine-learning10
artificial-intelligence10
nlp10
python10
fairseq10
bert10
data-pipelines9
data-pipeline9
roberta9
tokenizer8
tokenize8
testing8
Programming languages (2)
Github contributions (5)
pytorch/torchtune
Mar 2024 - Aug 2024
PyTorch native post-training library
Role in this project:
ML Engineer
ML EngineerContributions:310 reviews, 77 PRs, 157 pushes in 5 months
Contributions summary:Kartikay implemented Rotary Positional Embeddings (RoPE) and associated tests within the repository. Their primary focus was adding a specific neural network layer to the project related to LLM model architecture. Additionally, the user also added a .gitignore file.
facebookresearch/pytext
Apr 2019 - Jan 2021
A natural language modeling framework based on PyTorch
Role in this project:ML Engineer
ML EngineerContributions:19 commits, 17 PRs in 1 year 9 months
Contributions summary:Kartikay primarily contributed to the `pytext` repository by implementing and improving the data handling and processing pipelines for natural language modeling. Their work included adding new batch samplers like `ProbabilisticBatchSampler` for cross-lingual LM training and refining existing ones like `RoundRobinBatchSampler` and `EvalBatchSampler`. They addressed memory issues within metric reporting, improved tensorization processes and standardized the vocabulary creation. The user also added features like a `MultilingualTSVDataSource` to handle multi-lingual datasets, and integrated support for XLM-R models.
pytorchnlpbertmachine-learningnatural-language
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.