Data Scientist

About the company

Deriva.ai is building the first AI-powered platform that helps clinical geneticists classify genetic variants with the intuition of an experienced scientist. By fusing fine-tuned large-language models, multi-agent orchestration, and rich biomedical data, we cut analysis time from days to minutes—giving patients faster answers and clinicians deeper insight.

About You

You inhale fresh arXiv drops for breakfast and can’t resist porting a new loss function before the pre-print hits Twitter. You’re an applied Data Scientist who delights in turning bleeding-edge research—contrastive objectives, RLHF variants, retrieval tricks—into production impact. Prompt-crafting is a science, Airflow DAGs are your whiteboard sketches, and you keep a running list of “models to reproduce” the way others keep playlists.

Responsibilities

• End-to-end ML pipelines – design, implement, and maintain workflows for training, fine-tuning, and deploying models (LLMs, GNNs, RL loops).
• AI-agent prototyping – build and iterate on hierarchical agents with LangGraph/LangChain, crafting chain-of-thought prompts that actually think.
• Knowledge-graph integration – ingest and link entities to drive retrieval, reasoning, and explainability.
• Data orchestration – author Apache Airflow DAGs for ingestion, feature engineering, model retraining, and monitoring.
• Cloud-native ops – containerize models with Docker, then own their performance, reliability, and traceability in production.
• Cross-functional collaboration – partner with product and engineering teams to optimize inference speed, ensure compliance, and harden system robustness.

Qualifications

• Education & experience – Ph.D., or equivalent real-world track record in ML/Data Science.
• Core AI expertise – hands-on with LLM fine-tuning, RL (PPO / GRPO), contrastive-learning objectives; comfortable in PyTorch.
• Agent & prompt chops – practical experience with chain-of-thought prompting, RAG, and multi-agent frameworks (LangChain, LangGraph, etc.).
• Engineering fundamentals – strong Python; experience with Airflow, Docker, CI/CD, and at least one cloud provider.
• Self-starter mindset – you prototype fast, validate faster, and document as you go.

Preferred Extras

• Experience with graph databases.
• Familiarity with Kafka.

Why Deriva.ai

• Mission with impact – every model you ship helps clinicians deliver life-changing diagnoses faster.
• Tech playground – LLM fine-tuning, multi-agent orchestration, knowledge graphs, and a modern DevEx stack are your everyday gear.
• A-player crew & good vibes – collaborate with a sharp, supportive team that loves to teach, learn, and celebrate wins over craft beer or a killer playlist.
• Early-stage momentum – join while the rocket is still lifting off and shape how the startup grows from its early chapters.