Role overview

Qualifications

Foundation models: training, new architectures, RL, reward modeling, scaling
Evaluation: benchmarks, eval loops, quality measurement, LLM-as-judge, and failure analysis
PhD in ML/NLP or equivalent practical experience with evidence of research impact (publications, open-source work, or notable projects)
Full-stack research ownership: ability to frame questions, run experiments, write papers, and ship results

Responsibilities

Frame research questions, design and run experiments, analyze results, and ship outcomes (papers, models, or systems)
Own end-to-end evaluation as a first-class research product: build benchmarks, eval loops, metrics, and failure analysis
Collaborate with cross-functional teams, including US leadership and engineering, and communicate clearly in English
Produce visible impact by sharing work that moves the field forward and delivering results with a bias for impact over purely academic output

Key facts

Remote from: Germany
Full time
Senior (5-10 years)
English

Hard skills

Benchmarking Reinforcement Learning Foundation Engineering Natural Language Processing (NLP) Failure Analysis Reward Management Controlled Experiments Quality Assessment SQL Server Agent Controlled Experiments Multi-Agent Systems Real-Time Operating Systems Model-Based Design Scientific Writing Machine Learning Machine Learning Nested Loops Interdisciplinary Research Adaptive Learning

Other skills

Collaboration
Integrative Thinking
Communication
Dealing With Ambiguity
Problem Solving

About the company

Inworld AI

Inworld is a Character Engine for AI NPCs that goes beyond LLMs – and adds contextual awareness and multimodal expression of personality. Inworld uses advanced AI to build generative characters whose personalities, thoughts, memories, and behaviors are designed to mimic the deeply social nature of human interaction. The Inworld platform lets you create characters with personality and contextual awareness to keep them in-world and in character. Integrations make it easy for developers to deploy characters into immersive experiences, while scale and performance are optimized for real-time experiences. We are a team of experts that have pioneered conversational AI platforms and generative models at Google and DeepMind. We are continuing to build out our incredibly talented team, with experts in generative language models, emotions, speech synthesis, multimodal interaction, design, and 3D animation. Inworld is backed by top-tier investors including Section 32, Intel Capital, BITKRAFT Ventures, Kleiner Perkins, Founders Fund, First Spark Ventures, The Venture Reality Fund, CRV, Meta, Microsoft’s M12 fund, Micron Ventures, LG Technology Ventures, NTT Docomo Ventures, and SK Telecom Venture Capital. Inworld was one of six companies selected for the 2022 Disney Accelerator. Prominent angels include Twitch Co-Founder, Kevin Lin; Oculus Co-Founder, Nate Mitchell; Animoca Brands Co-Founder, Yat Siu; The Sandbox Co-Founder, Sebastien Borget and NaHCO3, the family office of Riot Games Co-Founder, Marc Merrill.

Company details

Company typeScaleup

Company size51 - 200

Links

Website LinkedIn See all jobs

Your match analysis

See how your profile stacks up against this role.

We compared the job requirements to your profile to show where you're strong and where you fall short.

Job description

About Inworld

Inworld is a product-oriented research lab of top AI researchers and engineers, developing best-in-class realtime multimodal models and the only realtime orchestration platform optimized for thousands of queries per second.

We’ve raised more than $125M from Lightspeed, Section 32, Kleiner Perkins, Microsoft’s M12 venture fund, Founders Fund, Meta and Stanford, among others. Our technology has powered experiences from companies such as NVIDIA, Microsoft Xbox, Niantic, Logitech Streamlabs, Wishroll, Little Umbrella and Bible Chat. We’ve also been recognized by CB Insights as one of the 100 most promising AI companies globally and have been named one of LinkedIn's Top 10 Startups in the USA.

Who We're Looking For

A year ago, reliably working agentic systems barely existed. Nobody has a decade of experience here. So we're not screening for a resume template — we're looking for strong people from varied backgrounds who learn fast, thrive in ambiguity, and can show us what they've built, broken, and understood.

Experience We Find Useful

You don't need all of this. But you need enough to make a case.

Foundation models: training, new architectures, RL, reward modeling, scaling
Evaluation: benchmarks, eval loops, quality measurement, LLM-as-judge, failure analysis
Frontier topics: multimodal models, agents, tool use, test-time compute, world models
Published research at ICML, ICLR, NeurIPS, EMNLP, ACL, or AAAI
PhD in ML/NLP — or equivalent practical experience you can point to
Public work: non-trivial AI side projects, interdisciplinary experiments, open-source contributions
Full-stack research ownership: you frame the question, run the experiments, write the paper, ship the result

If you learned through building, competitions, or collaborations outside academia — that counts. We care about evidence, not credentials.

Who Thrives Here

Pathfinders: You don’t need a roadmap to start walking; you’re comfortable picking a direction and building the map as you go.
Full-Cycle Researchers: You believe research isn't finished until it’s shipped. You have a bias for impact over purely academic output.
First-Principles Engineers: You don't just ship code; you obsess over the why. You’re the first to question an approach if you think there’s a better way to solve the core problem.
Mission Owners: You aren't satisfied with "the PM said so." You thrive on deep context and want to understand the fundamental logic behind every decision we make.

What Working Here Is Like

We hand you unclear problems and expect you to make them clear. We value researchers who say "I don't know yet" — and then design the experiment that finds out. We treat evaluation as a first-class research product, not a box to check before launch. Impact comes before publications though we support sharing work that moves the field forward. Your work should be visible. Flat structure, fast iterations, minimal process theater.

We don't need a cover letter. A link to something you've built tells us more.

Professional fluency in English (written and spoken) is required, as you will be collaborating daily with our US-based leadership and engineering teams.