10 OpenAI machine learning engineers
Model training is what nine of these OpenAI machine learning engineers describe in their current roles, from pretraining data and training throughput to post-training and reinforcement learning infrastructure; the tenth works on AI and ML for ChatGPT. Teams building their own training stack, or deciding where to spend on compute and data, can use them to sanity-check an architecture or a hiring plan.
Professionals to explore
Alex Paino
LinkedInExperience: Member of Technical Staff · OpenAI
$841 per 30-minute call
A Member of Technical Staff at OpenAI since 2019, Alex Paino manages the Pretraining Data team and previously worked on Robotics there. At Sift, Alex was Staff Software Engineer (2017-2019) and Machine Learning Engineer (2014-2017). How pretraining datasets are assembled and managed is the topic Alex is placed to discuss.
Is this you? Claim your profile and get paid for calls.
Hitesh Khandelwal
LinkedInExperience: Member of Technical Staff · OpenAI
$841 per 30-minute call
Hitesh Khandelwal became Member of Technical Staff at OpenAI in January 2026, working on agent infrastructure for RL training and research and for agentic apps in production. Hitesh was a Software Engineer at Facebook (2012-2026) and, in Meta Superintelligence Lab (MSL), founded and led the AI Inference Platform (IPnext). Inference platforms and agent infrastructure are what Hitesh can speak to.
Is this you? Claim your profile and get paid for calls.
John Zheng
LinkedInExperience: Member of Technical Staff · OpenAI
$841 per 30-minute call
Recommendation and applied ML in consumer products are the fit for John Zheng, Member of Technical Staff at OpenAI since June 2026, working on AI and ML for ChatGPT. In Engineering Leadership (AI ML) at Meta (2022-2026), John worked on Instagram growth modeling and LLMs for Instagram feed and reels content recommendation, and was a Machine Learning Engineer at Pinterest (2018-2022).
Is this you? Claim your profile and get paid for calls.
Łukasz Kondraciuk
LinkedInExperience: Member of Technical Staff · OpenAI
$841 per 30-minute call
Strawberry (o1) training systems lead is how Łukasz Kondraciuk describes the Member of Technical Staff role held at OpenAI since 2022. Łukasz was a Research Engineer at Huawei (2021-2022), interned at Google and on NVIDIA's Cuda Math Libraries, and is an ACM ICPC 2022 silver medalist, a fit for training systems for reasoning models.
Is this you? Claim your profile and get paid for calls.
Martin A.
LinkedInExperience: Member of Technical Staff · OpenAI
$841 per 30-minute call
In ChatGPT engineering since joining OpenAI as Member of Technical Staff in April 2026, Martin A. describes the work as ChatGPT and Codex and scaling and optimizing AI training infrastructure. Martin was a Software Engineer at LinkedIn (2024-2026) after internships at LinkedIn, BlackBerry and Hölmetrics, a fit for product-facing ML engineering and training infrastructure.
Is this you? Claim your profile and get paid for calls.
Praveen Kumar
LinkedInExperience: Senior Staff Machine Learning Engineer · OpenAI
$1,078 per 30-minute call
Post-training for frontier AI models is the work of Praveen Kumar as Senior Staff Machine Learning Engineer at OpenAI since 2025, with RLHF, LLMs, agentic AI and distributed training as stated areas. Praveen was Senior Staff Engineer at Google DeepMind (2022-2025), Staff Machine Learning Engineer at Meta (2018-2022) and Senior Software Engineer at Google (2008-2018), so post-training and RLHF questions fit.
Is this you? Claim your profile and get paid for calls.
Pritam Damania
LinkedInExperience: Member of Technical Staff · OpenAI
$841 per 30-minute call
Large training and RL systems design suits Pritam Damania, who joined OpenAI as Member of Technical Staff in October 2024 and builds large-scale training and RL systems for LLMs, world models, robotics and Sora, as a core contributor to Sora 2.0 and GPT-5.x. At Tesla (2023-2024), Pritam was Senior Staff Software Engineer and technical lead in the ML Infra team at Tesla AI.
Is this you? Claim your profile and get paid for calls.
Ricky X.
LinkedInExperience: Member of Technical Staff · OpenAI
$841 per 30-minute call
Ricky X. has worked on Research Platform Runtime, the training infrastructure, as Member of Technical Staff at OpenAI since 2025. Ricky was a Senior Software Engineer at Anyscale (2022-2025) and describes being a Ray Committer building distributed systems for ML and AI and a vllm contributor on LLM performance. Distributed training runtimes and open-source ML systems are where Ricky fits.
Is this you? Claim your profile and get paid for calls.
Trevor Cai
LinkedInExperience: Member of Technical Staff · OpenAI
$841 per 30-minute call
GPT-4 training throughput and vision execution were led by Trevor Cai, Member of Technical Staff at OpenAI since 2022. At Google DeepMind (2017-2022), Trevor was Tech Lead of DeepMind's LLM effort, built the infrastructure that trained Chinchilla and Gopher, and co-created Haiku, so training throughput and large-run infrastructure are Trevor's area.
Is this you? Claim your profile and get paid for calls.
Wenting Ye
LinkedInExperience: Member of Technical Staff · OpenAI
$841 per 30-minute call
Training @ Research is the description of the Member of Technical Staff role Wenting Ye took at OpenAI in August 2026. Wenting was Staff Research Engineer at Google DeepMind (2024-2026), Research Engineer at Character.ai (2024), Senior Machine Learning Engineer at TikTok (2022-2024) and Applied Scientist II at Amazon (2022), and can compare research engineering on model training at OpenAI with Google DeepMind and Character.ai.
Is this you? Claim your profile and get paid for calls.
Choose the right perspective
For post-training and RLHF, Praveen Kumar is the clearest match; for pretraining data, Alex Paino. Training systems and throughput questions fit Trevor Cai, Łukasz Kondraciuk, Pritam Damania or Ricky X., agent and RL infrastructure fits Hitesh Khandelwal, and applied ML in a consumer product fits John Zheng or Martin A.
Questions to take into the conversation
- 01How do OpenAI's training infrastructure engineers find and fix the bottlenecks that limit training throughput on a new large run?
- 02What changes in the engineering setup when a team moves from pretraining a model to post-training it with reinforcement learning?
- 03How do OpenAI engineers decide which training data pipeline problems are worth solving in-house versus with open-source tools like Ray?
About this directory
This is a professional research starting point based on business profile data retrieved on . Titles and companies reflect that source snapshot; roles marked former had ended by then. Check the linked profiles for current details. Inclusion does not imply Instant Expert membership or availability.