Machine Learning Engineer, Infra, AI for Drug Discovery
About The Position
Roche’s Research and Early Development organizations at Genentech (gRED) and Pharma (pRED) are leveraging advances in AI, data, and computational sciences to transform drug discovery and development. To maximize these opportunities, seamless data sharing and access to models across gRED and pRED are essential. The new Computational Sciences Center of Excellence (CoE) is a strategic, unified group focused on harnessing the transformative power of data and Artificial Intelligence (AI) to assist scientists in delivering innovative medicines. The AI for Drug Discovery (AI4DD) group (Prescient Design) is building the machine learning platforms that enable researchers and engineers to move models from experimentation into reliable scientific and production workflows. This role will contribute to our model-serving platform and the broader infrastructure required to make machine learning models easier to deploy, scale, observe, and safely incorporate into scientific and agentic workflows. The scope extends beyond LLM serving, encompassing a range of machine learning and scientific models, including real-time and batch inference workloads, GPU-backed services, agentic applications, and in-silico drug discovery workflows. This is a hands-on engineering role for someone who enjoys writing and shipping production software across application code, cloud infrastructure, Kubernetes, and distributed systems.
Requirements
- BS or MS in Computer Science, Engineering, or a related technical field, or equivalent practical experience.
- 3+ years of relevant industry experience in software engineering, infrastructure engineering, platform engineering, DevOps, MLOps, or a related area.
- Strong Python programming skills and experience building and shipping maintainable production software, services, automation, or developer tooling.
- A demonstrated interest in hands-on implementation and production software delivery.
- Experience designing, deploying, or operating cloud systems (preferably on AWS) using services such as EKS, EC2, S3, IAM, SQS, SNS, and CloudWatch.
- Experience with containers, Kubernetes, Helm, and IaC tools such as Terraform or Pulumi.
- Experience with CI/CD, Git-based development workflows, automated testing, and software release practices.
- Ability to troubleshoot complex systems using metrics, logs, traces, events, and observability tools such as Datadog, Prometheus, Grafana, or OpenTelemetry.
- Understanding of distributed-systems concepts such as concurrency, queuing, retries, timeouts, idempotency, backpressure, and failure recovery.
- Ability to gather requirements, communicate technical tradeoffs, and document systems for users and engineers with varied infrastructure experience.
- Demonstrated ability to independently deliver practical, incremental solutions while considering immediate needs and longer-term platform direction.
Nice To Haves
- Familiarity with model-serving or workflow-orchestration frameworks such as KServe, Triton, vLLM, Ray Serve, Prefect, or Dagster.
- Experience optimizing model startup time, request throughput, batching, autoscaling, or GPU utilization.
- Familiarity with model registries, experiment tracking, model evaluation, promotion workflows, or MLOps platforms.
- Experience building event-driven systems using queues, event buses, or workflow orchestrators.
- Familiarity with online and offline model evaluation, model-quality monitoring, data drift, or regression analysis.
- Experience supporting scientific computing, high-performance computing, distributed training, or large-scale data processing.
- Strong interest in the life sciences and drug discovery.
Responsibilities
- Design, implement, ship, and operate scalable model-serving infrastructure for machine learning, scientific, LLM, and agentic workloads.
- Help evolve our internal model deployment platform into a reliable, self-service platform for teams across the organization.
- Improve platform scalability and reliability, including scale-to-zero, faster model startup, workload isolation, traffic management, and reduction of request failures and latency bottlenecks.
- Build observability and operational tooling for model usage, latency, reliability, resource consumption, inference cost, bottlenecks, and service-level indicators.
- Improve the usability of model deployment by developing validated configuration interfaces, reusable deployment patterns, APIs, command-line tools, and documentation.
- Help converge real-time and batch inference workflows onto shared platform capabilities where appropriate.
- Contribute to model lifecycle management infrastructure, including model registration and versioning, evaluation, promotion and release gates, monitoring, environment progression, and rollback.
- Build event-driven integrations that connect model publication, evaluation, promotion, deployment, and retraining workflows.
- Build consistent metrics and evaluation signals for understanding model cost, quality, reliability, and fitness for downstream workflows.
- Partner with machine learning, data, scientific, and platform teams to translate requirements into maintainable solutions and remove infrastructure bottlenecks.
- Own workstreams from design through implementation and production support, using strong software-engineering practices including testing, reviews, documentation, and incremental delivery.
Benefits
- Discretionary annual bonus may be available based on individual and Company performance.
- Benefits detailed at the link provided below.
Stand Out From the Crowd
Upload your resume and get instant feedback on how well it matches this job.
What This Job Offers
Job Type
Full-time
Career Level
Mid Level
Education Level
Associate degree
