Language AI Solutions Specialist

Company OverviewAt Keywords, we are using our passion for games, technology and media to create a global services platform for video games and beyond. Our aim is to become the “go to” provider of technical services. We enable leading content creators and publishers to leverage our expertise and capacity across the lifecycle of interactive content. In so doing we empower our clients to remain lean and agile, and to focus on creating the most engaging experiences. Keywords is trusted and relied upon by many of the world’s leading video game companies to work alongside them during concept, development and live operations by leveraging the breadth and depth of our industry leading service lines every step of the way. www.keywordsstudios.com Role OverviewAs a Language AI Solutions Specialist, you will bridge the gap between traditional localization and cutting-edge AI. You will be responsible for the end-to-end lifecycle of translation systems—from legacy Neural MT to modern LLM-based solutions. Your mission is to architect, tune, and deploy high-performance language models that drive global scalability while maintaining uncompromising linguistic quality. Key Deliverables• Hybrid Engine Deployment: Design, test, and certify Neural MT systems (KantanMT) and fine-tuned LLMs tailored to client-specific domains. • Performance Analytics: Deliver comprehensive quality reports, including Predictive TER scores, MQM-DQF/COMET/G-Eval metrics, and Human-in-the-Loop (HITL) evaluations. • Production Integration: Working closely with internal stakeholders (Development, Production, etc.) to oversee the seamless transition of validated models into live production environments.

Core Responsibilities1. AI Model Tuning & Optimization • Fine-Tuning & Hyperparameters: Execute supervised fine-tuning (SFT) on LLMs to align output with Client/Project specific style, voice, and domain-specific terminology. • RAG Corpus Management: Architect and maintain high-quality corpora for Retrieval-Augmented Generation, ensuring that vector databases are populated with clean, relevant, and deduplicated data. • Data Engineering: Perform advanced data analysis, cleansing, and data generation to prepare clean datasets for model training and tuning. • Output Verification: Utilize native-level linguistic expertise to perform deep-dive audits of NMT and LLM outputs, identifying nuanced errors such as hallucinations, cultural insensitivity, or tone inconsistency that automated metrics might miss.

2. Linguistic Asset Engineering • Modernization: Manage and optimize Translation Memories (TMs) and Termbases to serve as the foundational "ground truth" for AI training. • Quality Control: Implement and evolve Machine Translation Post-Editing (MTPE) workflows, integrating AI-assisted quality estimation (QE) to reduce manual overhead.

3. Analytics & Productivity Insights • Impact Metrics: Gather and analyze post-deployment data, focusing on Edit Distance, time-to-market reduction, and cost-per-word efficiency. • Stakeholder Reporting: Translate complex technical data into actionable insights for Program Managers and clients, recommending rebuilds or retraining cycles where necessary.

4. Operations & Troubleshooting • Technical Support: Serve as the escalation point for engine failures, API latency issues, or unexpected model "hallucinations.". • Process Documentation: Create "Living Documentation" for evolving AI workflows to ensure team-wide alignment and scalability.