Researcher at Inworld

Setup: Remote - Serbia
Type: Full-time
Posted: Mar 18, 2026

Description

Inworld is seeking a researcher to join their team. The ideal candidate will have experience in foundation models, evaluation methods, frontier topics like multimodal models and agents, and published research. While a PhD in ML/NLP is preferred, practical experience is also considered. Candidates should demonstrate strong problem-solving abilities, a focus on impactful results, and a commitment to understanding the underlying principles of solutions.

Required Skills

Foundation Models
Reinforcement Learning
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

Additional Information

Required Previous Experiences: Experience in foundation models, including training, architecture development, reinforcement learning, reward modeling, and scaling.
Experience in evaluation methodologies, such as benchmarking, evaluating loops, quality measurement, using large language models as judges, and failure analysis.
Published research in conferences such as ICML, ICLR, NeurIPS, EMNLP, ACL, or AAAI.
Experience in public work, including non-trivial AI side projects, interdisciplinary experiments, and open-source contributions.
Ability to own full-stack research, encompassing framing questions, conducting experiments, writing papers, and implementing results.
Preferred Qualifications: Comfortable with exploring directions without a predefined roadmap.
Focus on delivering impactful results rather than solely academic outputs.
Commitment to questioning approaches to find better ways to solve core problems.
Desire to understand the fundamental logic behind decisions, not just following instructions.