Staff Data Scientist (Python / LLM / MLOps)

About Us

Gramian Consultancy is a boutique consultancy specializing in IT professional services and engineering talent solutions. With a strong background in software engineering and leadership, we help companies build high-performing teams by matching them with professionals who truly fit their needs.

This opening is on behalf of one of our clients, and we’ll work closely with you to make the process clear and straightforward.

About the company and the role

Our client is a fast-growing AI startup headquartered in the San Francisco. They are building smart data tools that make large language models (LLMs) more reliable and useful in real-world applications. Their platform helps turn messy, unstructured data into structured knowledge, making it easier for teams to build AI systems that are accurate and consistent. One of their key focus areas is the legal industry, where their product helps law firms move faster by spotting risks early, managing plaintiffs at scale, and applying AI-native tools to streamline discovery.

About the role

They are looking for a Data Scientist (Python / LLM / MLOps) with 2–4 years of experience. In this role, you’ll work at the intersection of data science, machine learning, and production systems, helping transform messy real-world data into structured insights that power AI-driven legal products.

This is a hands-on role where you’ll explore, model, and deploy at startup speed — building solutions that improve the reliability of large language models and make them more useful in high-stakes, real-world applications.

The position requires working in PST time zone.

Responsibilities

• Data Exploration (EDA): Quickly profile raw JSON/CSV/HTML, spot outliers, and uncover actionable patterns without waiting for detailed specs.
• Model Development: Train and tune classification, ranking, and LLM-based models to continually improve precision and recall.
• API Deployment: Package models into FastAPI services, validate schemas with Pydantic, and integrate with CI/CD pipelines.
• MLOps: Monitor model drift, schedule batch evaluations, and build lightweight tests to ensure production reliability.
• Communication: Deliver clear notebooks, dashboards, and concise demos that legal subject-matter experts can understand and act on.
• Startup Mindset: Proactively identify and fix data, labeling, or infrastructure gaps — thriving in an environment where ambiguity is the default.

Requirements

Must-Haves (all required):

• 2–4 years of Python data science experience.
• Recent experience at a fast-growing startup
• Hands-on experience training/tuning ML models, including LLMs.
• Strong knowledge of Python data stack (pandas/Polars, PyTorch or TensorFlow).
• Proven experience building and deploying FastAPI services with Pydantic.

Nice-to-Haves (meet at least 70%):

• Experience with web scraping (Scrapy, Playwright) or working with public datasets (PACER, NHTSA, FDA).
• Familiarity with vector databases (Qdrant, pgvector) and prompt-engineering.
• Exposure to regulated-data frameworks (SOC 2, HIPAA).
• Strong CI/CD skills (GitHub Actions).
• Ability to document and demo work clearly and proactively.

Benefits

• Equity
• Remote work