Top AI Research Jobs in Austin, TX

Conduct applied AI safety research using white-box methods such as activation explainers, steering, attribution, and model-internal interventions. Evaluate methods on realistic, large-scale models and agentic coding tasks, develop practical monitors and interventions, publish findings, propose research directions, and engage with researchers, frontier AI developers, governments, and the broader AI alignment community.

Co-develop an EIC Pathfinder proposal focused on trustworthy cognitive AI, including causal reasoning, abstraction, planning under uncertainty, methodology, related work, and evaluation design. The researcher will help build a working TRL 3/4 system, develop benchmarks, and collaborate closely with a startup team through submission and subsequent project activities.

Leads research on the economic impact of AI and technology disruption across markets, business models, and value chains. Synthesizes economic analysis, quantitative evidence, financial statements, market intelligence, and emerging signals into Gartner thought leadership and executive recommendations. The role includes client engagement, presentations, media interactions, content development, peer review, sales support, mentoring, and shaping research agendas.

Conduct research on frontier AI systems and develop datasets, experiments, prototypes, tooling, evaluation frameworks, and reinforcement learning environments. Investigate synthetic and agentic data generation, post-training, model understanding, benchmarks, and evaluation. Collaborate with research, engineering, product, and operations teams to translate findings into scalable AI applications, while contributing to technical publications, open-source work, mentorship, and the broader AI research community.

Leads development of agent evaluation systems that assess live software agents using trace data, rubrics, benchmarks, and LLM-as-a-judge techniques. Owns the evaluation platform’s full stack, creates reusable workflows and integrations, tracks emerging evaluation methods, and collaborates with data science teams. Manages and mentors engineers while promoting evaluation practices across agent-focused products.

Research and prototype transformer-based and graph-based foundation models using customer, transaction, and merchant data. The intern will design and train models, run rigorous experiments, evaluate results, and develop working prototypes. Responsibilities include collaborating with Product, Engineering, Analytics, and Applied Science teams to improve customer experience, operational efficiency, and business outcomes.

Conduct industry-focused AI research on autonomous agents and world models. Develop models involving large language models, active inference, and optimal policies, including architecture experimentation, fine-tuning, deep-learning agent development, and action optimization. Candidates will work on real-world enterprise problems while pursuing cutting-edge research during a fixed-term three- or six-month fellowship.

Conduct independent research on frontier AI evaluation and auditing. Design rigorous evaluations, audit methods, and reproducible experimental pipelines; analyze model capabilities, risks, robustness, and alignment-relevant behavior. Produce audit reports, recommendations, publications, tools, and presentations while collaborating with technical, policy, engineering, and standards teams. The role requires careful handling of confidential information and may involve pilot audits with frontier AI companies.

Join Arootah's advisor network to provide fractional Chief AI Officer / AI Strategy leadership to hedge funds, private equity firms, and family offices. Advise on AI initiatives, strategy, governance, vendor evaluation, workflow automation, change management, and implementation roadmaps. Engage on project-based assignments; client origination handled by Arootah.

Evaluate and improve clinical AI systems through grading-criteria development, dialogue evaluation, clinical reasoning annotation, output review, guideline authoring, and difficult-case writing. The role requires physicians to assess accuracy, safety, completeness, clinical reasoning, and escalation guidance without providing direct patient care. Work is assigned across shared clinical AI workstreams and requires at least 20 hours weekly.