Research Engineer, AI for Science: OpenAI

This is a "moonshot" role within OpenAI. You are not optimizing ChatGPT for general consumers; you are building the "next great scientific instrument." The AI for Science team operates with a specific mandate: to prove that frontier models (like o1 or GPT-5) can perform "real science"—generating hypotheses, designing experiments, and accelerating discoveries in fields like biology, physics, and chemistry.

You will sit at the exact intersection of Software Engineering and Research. Your job is to take the experimental ideas of research scientists and build the massive-scale distributed systems required to make them work reliably on thousands of GPUs.

Key Responsibilities
Scale: Design and implement distributed machine learning systems. You aren't just training models; you are building the infrastructure that allows models to ingest massive scientific datasets (e.g., genomic sequences, molecular structures).

Optimization: Write high-performance, robust ML code. This likely involves optimizing training loops, data loading pipelines, and inference engines.

Translation: Act as the bridge between pure research and engineering. You must translate a researcher's mathematical concept into scalable, bug-free code.

Qualifications
Engineering First: Unlike "Research Scientist" roles that require a PhD and publication history, this is a "Research Engineer" role. They prioritize strong coding skills and distributed systems experience over academic pedigree.

HPC (High-Performance Computing): Experience with large-scale compute (Kubernetes, Ray, MPI, NCCL) is critical.

Scientific Curiosity: A "Nice to have" is a genuine interest in science (biology/physics). If you have a background in Scientific Computing (e.g., simulations, fluid dynamics) alongside your CS skills, you are a top-tier candidate.

Demographics & Representation (2024-2025 Estimates)
As a private company, OpenAI does not publicly file detailed EEO-1 diversity reports like public tech giants (e.g., Adobe). However, recent third-party analyses and partial disclosures provide the following snapshots:

Gender: Approximately 44% of OpenAI's total workforce identifies as women or non-binary (Source: ElectroIQ / SQ Magazine Analysis 2025).

Department Split: The company is roughly 56% Engineering/Research and 44% Operations/Go-To-Market.

Growth Context: The workforce has expanded explosively, from ~770 employees in late 2023 to over 3,500 in late 2024.

Racial/Ethnic Data: Specific internal breakdowns by race are not publicly disclosed by OpenAI.

Strategic Context: Why this role exists
OpenAI is moving from "Generative AI" (writing text/images) to "Reasoning AI" (solving complex problems).

The Competitor: Google DeepMind has dominated the "AI for Science" niche with AlphaFold (protein structure prediction) and AlphaGeometry.

The Goal: OpenAI wants to prove its models are not just chatbots, but reasoning engines capable of Nobel-Prize-level scientific breakthroughs. You are being hired to build the engine that drives this competition.