Research Engineer, Scientific Data & Workflow Infrastructure

Lila Sciences•Cambridge, MA

1d•$148,000 - $240,000

About The Position

Lila is building an operating system for science, powered by scientific superintelligence — generating hypotheses, designing and running experiments, and learning from new data in real time to speed up discovery in fields like therapeutics and advanced materials. The Molecular Intelligence team builds ML and physics-based infrastructure for drug discovery: biophysics workflows, computational chemistry tools, cofolding models, low-data learning systems, simulation workflows, and agent-usable scientific pipelines. This role sits embedded with scientists, working closely with Lila’s platform teams on the storage, compute, and orchestration layers beneath it. You’ll build the provenance system that connects every scientific input, code version, model, run, and result produced across Lila’s simulation, ML, and agent-driven workflows — so a scientist can find a result months later, trace exactly how it was produced, and reuse it without reconstructing it from logs.

Requirements

Strong Python engineering skills with a track record of designing and shipping maintainable systems.
Experience with data modeling — schemas, metadata, identifiers, versioning for data that changes over time.
Experience designing APIs and building libraries that other engineers or scientists depend on.
Experience supporting research workflows (ML experiments, simulations, data pipelines) with a clear sense of what makes results reproducible.

Nice To Haves

Familiarity with chemistry, computational biophysics, molecular simulation, or drug discovery workflows.
Experience with experiment tracking / data lineage tools (MLflow, Weights & Biases, OpenLineage, DVC).
Experience with workflow orchestration (Flyte, Ray, Argo, Airflow, Temporal, Dagster).
Experience with Kubernetes, containers, and CI/CD at scale.
GPU computing experience, ideally at hundreds-to-thousands-of-GPU scale.
Experience building tools for LLM agents or automated research workflows.

Responsibilities

Build shared provenance infrastructure — a queryable, trusted record connecting inputs, code/model versions, runs, and outputs across simulation, ML, and agent workflows.
Develop data models and metadata schemas for runs, artifacts, and results that work across heterogeneous scientific tools.
Create reusable Python libraries and APIs that let tools, pipelines, and AI agents write to and read from shared storage and metadata systems.
Structure records from agent activity — turning ephemeral agent traces into durable, queryable data.
Transition production-grade tooling from research prototypes — taking scientific workflows from one-off scripts to scalable, maintainable services.
Establish engineering conventions (CI, testing, packaging, release practices, documentation) that hold up across local, cloud, and GPU compute environments.