AI Research Jobs in 2026: OpenAI and Scale AI Are Tied at #1—and Capital One Outranks Google
Most candidates targeting AI Research roles optimize for one company: OpenAI. The logic is understandable — OpenAI publishes the most cited papers, has the most recognizable brand in AI, and is headquartered in the city that AI researchers have flocked to for a decade.
HireIndex’s current index of 1,691 open AI/ML roles across 757 companies (week of July 27, 2026) tells a more distributed story. AI Research titles account for 116 roles — 6.9% of the total market. And the employer leading the category isn’t OpenAI alone.
OpenAI and Scale AI are exactly tied at 12 open AI Research roles each. The frontier lab and the AI services company are co-leaders in research headcount by one measure that actually matters: open positions right now.
The number behind #3 is more surprising. Capital One has 6 open AI Research roles. Google has 5. A bank is outpacing one of the world’s largest AI research organizations in live job postings.
How many AI Research jobs are open right now?
116 in the July 27 index. Here’s the full employer ranking:
| Company | Open AI Research Roles |
|---|---|
| OpenAI | 12 |
| Scale AI | 12 |
| Capital One | 6 |
| Waabi | 6 |
| 5 | |
| Databricks | 3 |
| Snorkel AI | 3 |
| Apple | 2 |
| Cohere | 2 |
| G-Research | 2 |
| JetBrains | 2 |
| Pika | 2 |
| PointClickCare | 2 |
| Spotify | 2 |
For context: Machine Learning Engineer accounts for 337 roles in the same index — roughly 2.9x the AI Research total. Research roles are a real market, not a rounding error, but they’re meaningfully narrower than the applied engineering categories. Candidates who have only been applying to ML Engineer roles may be missing a parallel category where the competition dynamics are different.
What makes OpenAI’s and Scale AI’s research agendas different?
Both companies are at 12 open research roles, but the work they’re hiring for is almost entirely distinct.
OpenAI’s research openings cluster around internal product and capability priorities:
- Research Engineer / Scientist — Personal AGI (Post Training, Personality, Model Behavior)
- Research Engineer — RL/Reasoning
- Researcher, Trustworthy AI
- Researcher, Health AI
- Research Engineer — Multimodal Agents
- Security Researcher, Agentic AI Threats
The thread running through these is alignment-adjacent work inside a company shipping consumer AI products at scale. OpenAI isn’t just hiring researchers to publish papers — it’s hiring people to make its models safer, more useful, and harder to exploit.
Scale AI’s research list reads differently:
- Research Scientist, Agent Robustness
- Research Scientist, AI Controls and Monitoring
- Research Scientist, Frontier Risk Evaluations
- ML Research Engineer, Agent Post-Training
- ML Research Engineer, Agent Data Foundation
Scale AI’s research hiring is almost entirely focused on evaluating, red-teaming, and improving agentic AI systems. The company has positioned itself as the evaluations and data layer for the frontier model era — its research function is building the infrastructure to assess whether AI systems are safe and reliable, not building the models themselves.
These are not interchangeable hiring pipelines. Candidates who focus on mechanistic interpretability and alignment fit the OpenAI path. Candidates who focus on evaluation methodology and safety benchmarks fit Scale AI’s.
What is Capital One actually hiring AI researchers to do?
Six mid-level “Applied Researcher” roles, all inside an “AI Foundations” team:
- Applied Researcher I — AI Foundations, Recommendation Systems, Personalization, Reinforcement Learning
- Applied Researcher I — AI Foundations, LLM Core and Agentic AI
- Applied Researcher I — AI Foundations, LLM Customization, Finetuning, Reinforcement Learning
- Applied Researcher II — AI Foundations
- Applied Researcher II — AI Foundations, LLM Core and Agentic AI
Capital One is building LLM fine-tuning and agentic AI capabilities in-house — not just deploying vendor models, but doing the foundational work of adapting them to financial services use cases. The RL and recommendation systems focus echoes the company’s earlier pattern with the LLM Engineer category, where it posted more roles than any AI lab.
This is the same bank that the LLM Engineer market data flagged in May 2026: Capital One has been quietly building one of the most technically serious AI engineering functions in financial services. The research roles confirm this is a deliberate internal capability, not a vendor integration project.
Where are AI Research jobs located?
This is where AI Research diverges sharply from most other AI/ML categories.
| City | Open AI Research Roles |
|---|---|
| Remote | 71 |
| San Francisco | 17 |
| London | 12 |
| New York | 8 |
| Toronto | 6 |
| Seattle | 1 |
| Boston | 1 |
61% of AI Research roles are Remote. That’s the highest remote concentration of any skill category in the index — significantly above the overall market average. The reason is structural: research work tends to produce verifiable outputs (papers, model improvements, benchmark results) that make location-independent accountability tractable in a way that production engineering often isn’t.
San Francisco’s 17 research roles are almost entirely OpenAI. London’s 12 are more distributed: Cohere, G-Research (a quantitative finance firm with genuine AI research depth), and spillover from companies with UK offices. Toronto’s 6 are similarly spread, including Waabi — the autonomous vehicle company — and a cluster of Canadian AI labs.
What seniority level are AI Research roles?
This is counterintuitive. The perception of AI Research as a field for the most senior researchers with decades of experience doesn’t match what’s in the index.
| Seniority | Count | Share |
|---|---|---|
| Mid-level | 89 | 77% |
| Senior | 15 | 13% |
| Senior / Principal | 9 | 8% |
| Entry-level | 2 | 2% |
| Director / Leadership | 1 | 1% |
77% of open AI Research roles are classified as mid-level. The two exceptions in the entry-level category are both PhD internships (Databricks and Waabi), not full-time hires.
The mid-level classification maps to recent PhD graduates and early-career researchers with 2–5 years of post-graduate experience, not the tenured professor-equivalent profiles that dominate AI research public discussion. The market is open to people earlier in their careers than most candidates assume.
The lean toward mid-level also reflects the maturation of the field. Five years ago, AI Research hiring was dominated by a handful of labs chasing senior academics. The market has scaled — more companies are building research functions, at earlier career stages, across more focus areas.
Explore all open AI Research roles and compare to the Machine Learning Engineer market on HireIndex.