Research Scientist

Pocket Entertainment is building the world’s largest AI Entertainment Platform, reimagining how stories are created, distributed, localized, and experienced globally. AI is core to our technology and product ecosystem. We are building across Generative AI, LLMs, multimodal AI, voice, conversational systems, localization, content intelligence, and agentic workflows to transform how entertainment is created and consumed.About the role We are looking for an experienced Research Scientist in Generative AI to work on a broad range of applied AI problems across Pocket Entertainment. This is not a role focused on a single model or modality. You could work on problems spanning LLMs, conversational AI, vision, multimodal models, content generation, adaptation and localization, AI agents, retrieval, personalization, and evaluation. You will take ideas from research and experimentation through to production, working on AI systems that directly impact how content is created, adapted, produced, and experienced by users globally. We are particularly interested in candidates who combine strong applied research skills with the ability to build and ship AI systems end-to-end.What you'll work on You'll work at the frontier taking research ideas into production, building systems that are both technically rigorous and genuinely useful at scale. The best AI products emerge when research, engineering, and creative judgment converge.Research, prototype, and ship state-of-the-art generative AI and multimodal systems from proof of concept to productionBuild for long-form content at scale: narrative reasoning, character consistency, story structure, and cross-lingual adaptation that preserves meaning and voiceDevelop conversational AI and agentic systems spanning multi-turn dialogue, memory, personalization, tool use, and complex workflow orchestrationDesign retrieval and knowledge systems using RAG, semantic search, vector databases, and knowledge graphsTrain and fine-tune models using SFT, RLHF, DPO, LoRA, and reinforcement learning and build the evaluation frameworks to measure what mattersOptimize for production: latency, throughput, reliability, and inference cost, while continuously improving quality from real-world data and model failuresCollaborate across research, engineering, product, design, and creative and contribute to publications at leading AI/ML venues where the work warrants itAreas you may work across Generative AI & LLMs: Long-form generation, reasoning, fine-tuning, and evaluationConversational AI & agents: AI characters, memory, tool use, and workflow orchestrationMultimodal & vision: Image and video understanding, generation, and creative workflowsLocalization & adaptation: Multilingual LLMs, cultural adaptation, and quality evaluationAI content & storytelling: Narrative generation, creator copilots, and story intelligenceRetrieval & knowledge systems: RAG, embeddings, knowledge graphs, and large-scale retrieval Voice & audio AI: Expressive TTS, speech understanding, and audio intelligenceWhat we're looking forMaster's or PhD in Computer Science, Machine Learning, AI, NLP, Computer Vision, or a related field or equivalent depth gained through industry experienceStrong Python skills and a solid foundation in modern deep learning and generative AI architectures, with hands-on experience training, fine-tuning, and shipping systems end-to-endExperience across one or more of: LLMs, conversational AI, multimodal AI, computer vision, machine translation, speech, or generative models with PyTorch or TensorFlow as your primary toolkitHands-on experience with model adaptation techniques SFT, RLHF, DPO, LoRA, or reinforcement learning and a rigorous approach to evaluation, including synthetic data pipelines and LLM-as-a-judge systemsExperience with retrieval and knowledge systems RAG, vector databases, knowledge graphs, or large-scale search and rankingExperience with long-form content generation, narrative AI, multilingual models, or culturally aware adaptationBackground in personalization, recommendation systems, or consumer-facing AI products at scaleComfortable moving fluidly between research and production translating advances in AI into measurable improvements for real usersPublications at leading AI/ML/NLP/CV conferences are a strong plus