Senior AI Researcher

At Phrase, we help open the door to global business by providing the world's leading Language Intelligence Platform.

The Phrase Platform combines AI, agentic orchestration, and a headless, API-first architecture in one composable system. Beyond translation, it orchestrates and adapts content to culture, audience, channel, brand voice, and intended outcome in any language, for every audience. It applies the context that makes content perform in every market: quality standards, glossaries, prior translations, and cultural nuance. Every team, in every region, can ship content that is on-brand, on-point, and ready for any audience.

Phrase gives enterprises the intelligence to automate workflows, the freedom to connect their own tools and engines, and the control to govern global content at scale.

With a global team based in our offices, and remote colleagues across Europe, the UK, the US, and the APAC region, Phrase offers an international environment built on collaboration, innovation, and shared purpose.

Phrase's AI Research team builds, trains, and evaluates the proprietary translation models and agentic workflows that power the platform's language quality — live systems serving production traffic, not proof-of-concept work. This role takes ownership of a significant research workstream within that portfolio, extending Phrase's live systems and shaping the direction of the team's wider research. You'll work across fine-tuning and instruction-tuning LLMs for domain-specific machine translation, designing evaluation frameworks that combine automatic metrics and LLM-as-judge approaches, and building agentic workflows for multi-step reasoning and tool use. It's built for someone with strong hands-on research experience who wants real scope to lead a body of work, mentor others, and grow their technical influence over time.

What you'll be responsible for:

• Design, train, and evaluate LLMs for translation and language-quality tasks, using fine-tuning techniques such as LoRA and DPO and instruction-tuned models.

• Design and implement evaluation frameworks for translation quality, combining automatic metrics (BLEU, TER, ChrF, MQM, COMET), LLM-as-judge approaches, and hybrid pipelines.

• Build and evaluate agentic workflows for NLP problems, spanning multi-step reasoning, tool use, and structured output generation.

• Own a significant workstream within the team's model-development programme, from data and training through evaluation and production integration.

• Design and run reproducible experiments, documenting results and decisions clearly.

• Work with Product, Engineering, and Solutions to turn research findings into shippable capabilities.

• Benchmark frontier models — OpenAI, Gemini, Claude, and open-source — against internal baselines.

• Evolve the team's quality-evaluation systems, including style-guide integration and customer-led, outcome-oriented metrics.

• Shape decisions on active learning, feedback-loop design, and data-pipeline strategy for continuous model improvement.

• Mentor junior researchers, review PRs and research write-ups, and contribute to reading groups and deep dives.

• Contribute to hiring, including take-home review and technical interviews.

• Represent your work in cross-functional forums with Product and Engineering.

What you need:

• Strong hands-on experience fine-tuning LLMs or large transformer-based models; familiarity with pre-training and preference-tuning (RLHF/DPO) is desirable.

• Solid NLP background, with demonstrable expertise in at least one of: machine translation, MT evaluation, information extraction, text generation, search optimization, or reinforcement learning.

• Ability to design, build, and evaluate agentic systems, including tool-calling agents, multi-agent pipelines, or LLM-orchestrated workflows.

• Experience designing and running evaluation frameworks, including automatic metrics and human-in-the-loop evaluation design.

• Strong Python skills and experience with ML/NLP libraries such as Hugging Face or PyTorch.