Forward Deployed Research Scientist
Job description
About the role
You will engage directly with frontier laboratory research teams as a technical peer during client scoping meetings, challenging assumptions about data requirements and shaping project specifications based on a scientific understanding of how data composition affects model outcomes. You will develop a deep scientific understanding of each client engagement, building working models of their architecture, training methodology, and target capabilities to reason empirically about data strategy and identify risks early. You will run ablation studies and fine-tune open-weight models to validate that delivered data improves model performance and catch problems before clients encounter them. You will consult on workflow and quality systems, reviewing annotation schemas, task designs, and quality controls in partnership with the Human Data Operations team to ensure alignment with research objectives. You will operate on client engagement timescales of days to weeks, managing multiple projects simultaneously while maintaining scientific rigor and rapid execution. You will translate empirical findings from fine-tuning experiments into actionable recommendations that guide data collection and labeling strategy for complex AI development challenges. You will collaborate closely with the Applied Research team to transform client-grounded insights into publishable work and ensure that research discoveries are integrated into production data pipelines. You will own the full lifecycle of data-centric research embedded within client engagements, from initial scoping through validation and knowledge transfer.
Key facts
What you'll do
- Engage directly with frontier laboratory research teams as a technical peer during client scoping meetings, contributing to methodology discussions and challenging assumptions about data requirements in real time.
- Develop detailed scientific understanding of client AI architectures, training methodologies, and target capabilities to evaluate how data composition influences model performance and strategic decision-making.
- Run controlled ablation studies and fine-tune open-weight models using client data and carefully selected proxy data to empirically measure the impact of data strategies on model outcomes.
- Consult on annotation workflow and quality systems, reviewing task design, schema definitions, and quality control mechanisms in collaboration with Human Data Operations to ensure rigorous data collection.
- Translate empirical findings from fine-tuning experiments and model evaluations into concrete recommendations that inform data collection priorities and project specifications for demanding AI research programs.
- Operate at the intersection of research science and client delivery, embedding research capability directly into engagements that drive business value and support frontier model development.
- Partner with the Applied Research team to synthesize insights from client projects into published work, turning grounded observations and experimental results into contributions to the broader AI research community.
- Manage multiple client engagements simultaneously, adapting to shifting priorities and tight timelines while maintaining scientific rigor, clear documentation, and consistent communication.
- Design and execute rapid experimentation plans that validate data hypotheses, using metrics from model performance and empirical analysis to guide iterative improvements in data strategy.
- Collaborate with cross-functional stakeholders to align data-centric research objectives with client goals, ensuring that deliverables are actionable, measurable, and technically sound.
Requirements
- You hold a PhD or equivalent practical experience in a relevant research field, with a strong record of scientific work that demonstrates deep methodological understanding.
- You possess proven experience running ablation studies and fine-tuning models, with a portfolio that shows how empirical results have influenced data-driven decision-making.
- You have direct experience working with open-weight models and understand the practical implications of model architecture and training dynamics for data strategy.
- You have a history of collaborating with research teams at major AI laboratories, navigating complex technical discussions and contributing at the level of scientific peers.
- You are capable of building clear mental models of complex AI systems and using them to reason about data requirements, risk factors, and expected outcomes in client engagements.
- You demonstrate mastery in designing and executing experiments that measure the impact of data composition on model performance under realistic conditions.
- You communicate effectively with both technical and non-technical stakeholders, translating empirical findings into recommendations that non-specialists can act on without loss of rigor.
- You are comfortable operating in a fast-paced environment where responsibilities expand quickly and success is measured by concrete impact on client research outcomes.
Practical notes
The role operates at the pace of client engagements, with work structured around days to weeks rather than long-term theoretical projects. Travel requirements are determined by client needs and project scopes as defined in source materials. Visa sponsorship considerations and specific hour allocations are to be reviewed in the source documentation associated with this position. Candidates must meet all eligibility criteria outlined in the requirements section without exception, as these form the baseline for scientific contribution and collaboration within the Alignerr team. The environment emphasizes ownership, rapid iteration, and continuous learning, ensuring that contributors remain at the forefront of data-centric AI research while delivering measurable results for clients and the broader research community.