Machine Learning Infrastructure Engineer

Job description

About the Team At Genesis Molecular AI, we're a tight-knit team of proven deep learning researchers, software engineers, and drug discovery pioneers. Our shared mission is nothing short of revolutionary: to forge the next generation of AI foundation models that unlock new therapies for patients with severe diseases. We conduct fundamental research at the intersection of machine learning, physics, and computational chemistry, pushing the boundaries of each field. You will work alongside machine learning researchers, computational scientists, and engineers to build the infrastructure that transforms massive, heterogeneous protein and chemical datasets into reliable, high-performance inputs for our models and drug discovery workflows. About the Role We are looking for a Machine Learning Infrastructure Engineer to build the data and orchestration systems that power molecular AI at Genesis. Drug discovery creates unusual infrastructure challenges: our workflows combine computational chemistry, structural biology, machine learning, and large-scale data processing in ways that don't map neatly onto traditional ETL systems. To meet these requirements, we have built our own pipeline orchestration framework—and we want an engineer excited to push it much further. This role is deliberately split between building the platform and using it. Roughly 50% of your time will be spent designing and improving our in-house orchestration framework: its DAG abstractions, execution engine, scheduling, caching, observability, and developer experience. The other 50% will be spent building and optimizing the protein, chemical, and ML data preprocessing pipelines that run on top of it. You'll work across both layers, finding opportunities to eliminate unnecessary computation, reduce latency, introduce lazy evaluation and caching, and make complex scientific workflows fast, reproducible, and easy for researchers to use. Your day-to-day work will span: - Design and evolve our in-house workflow orchestration framework, including DAG construction and execution, dependency management, scheduling, caching, retries, observability, and distributed execution. - Build and optimize large-scale preprocessing pipelines for protein structures, chemical datasets, simulations, and machine learning training data. - Profile end-to-end workflows and aggressively eliminate bottlenecks—from redundant I/O and serialization to unnecessary recomputation and poorly parallelized workloads. - Develop abstractions for lazy execution, incremental computation, intelligent caching, and artifact reuse so expensive scientific computations happen only when necessary. - Partner closely with ML researchers, computational chemists, and scientific software engineers to translate complex research workflows into scalable, reproducible computational pipelines. - Improve the developer experience for scientists and engineers authoring pipelines, making sophisticated distributed workflows intuitive to define, debug, monitor, and extend. - Make architectural decisions spanning local and distributed execution, storage, compute scheduling, data lineage, and reproducibility. - Use lessons from real-world scientific pipelines to continuously improve the orchestration platform itself—and use improvements to the platform to unlock faster, more ambitious scientific workflows. You are - Passionate about DAGs. You naturally think about computation as graphs of dependencies and care deeply about how work is scheduled, parallelized, cached, retried, and recomputed. - Impatient about latency. When a pipeline takes hours, your instinct is to understand exactly where the time went and systematically make it faster. - An evangelist for lazy execution and caching. You dislike unnecessary work and look for principled ways to avoid recomputation, move less data, and reuse intermediate results. - A strong systems engineer. You are comfortable reasoning across APIs, distributed systems, storage, serialization, concurrency, resource scheduling, and performance. - Hands-on with data-intensive systems. You have built production pipelines or infrastructure that processes large datasets reliably and efficiently. - Comfortable moving between framework and application layers. You're as interested in designing the orchestration primitive as you are in optimizing the pipeline built with it. - A pragmatic abstraction builder. You can identify the common pattern hiding underneath many specialized workflows without forcing every scientific problem into an overly generic framework. - A strong collaborator with researchers. You can understand an evolving scientific workflow, identify its computational structure, and turn it into robust infrastructure without slowing down experimentation. - Energized by unusual problems. You enjoy environments where off-the-shelf infrastructure gets you 80% of the way there—and the interesting work is designing the remaining 20%. Nice to haves - Experience building or operating large-scale ETL, data processing, or workflow systems such as Apache Spark, GCP Dataflow / Apache Beam, Flyte, Dagster, Airflow, Ray, or similar infrastructure. - Experience designing workflow engines, schedulers, DAG execution systems, build systems, or other dependency-driven computation frameworks. - Experience optimizing large-scale scientific, ML, protein, cheminformatics, or computational biology data pipelines. - Familiarity with Kubernetes and containerized compute environments, including deploying and operating distributed workloads across heterogeneous CPU and GPU resources. - Familiarity with Terraform or similar infrastructure-as-code tooling for provisioning and managing the underlying cloud, storage, networking, and compute resources that support large-scale pipelines. - Experience with cloud object storage, distributed compute environments, and high-performance or GPU-based computing. - Experience with content-addressable storage, incremental computation, data lineage, memoization, or cache invalidation at scale. - Familiarity with molecular data formats and tools such as RDKit, OpenEye, BioPython, molecular dynamics tooling, or structural biology pipelines. Compensation, Benefits, and Perks - Competitive compensation package that includes salary and equity. - Comprehensive health benefits: Medical, Dental, and Vision (covered 100% for the employees). - 401(k) plan. - Open (unlimited) PTO policy. - Free lunches and dinners at our offices. - Paid family leave (maternity and paternity). - Life and long- and short-term disability insurance. About Genesis Molecular AI Genesis Molecular AI http://genesis.ml is pioneering foundation models for molecular AI to unlock a new era of drug design and development. Our generative and predictive AI platform, GEMS (Genesis Exploration of Molecular Space), integrates AI and physics into industry-leading models to generate and optimize drug molecules, including the breakthrough generative diffusion model Pearl https://www.businesswire.com/news/home/20251028030745/en/Genesis-Molecular-AI-Unveils-Pearl-a-Field-Leading-Foundation-Model-that-Achieves-Unprecedented-Performance-in-Drug-Protein-Structure-Prediction for structure prediction. Genesis is backed by premier AI and life science investors, including a16z, NVIDIA, Rock Springs Capital, Menlo Ventures, T. Rowe Price, Fidelity, and Radical Ventures. Genesis has also signed category-leading AI-pharma deals, the most recent of which was a significant expansion with Incyte (see coverage in Forbes https://www.forbes.com/sites/innovationrx/2026/05/20/inside-incytes-120-million-ai-for-drug-development-deal/ and GEN https://www.genengnews.com/topics/artificial-intelligence/small-molecules-to-big-partnership-incyte-genesis-expand-ai-collaboration-to-1b/) with a total potential deal value of several billion dollars. Genesis is headquartered in San Mateo, CA, with a fully integrated laboratory in San Diego. We are proud to be an inclusive workplace and an Equal Opportunity Employer.