Chai Discovery

Software Engineer, Platform & Inference

San Francisco officefulltimemidAdded 6 days ago

About this role

Chai Discovery seeks a seasoned infrastructure engineer to build resilient, elegant systems powering AI-driven drug discovery. You'll design compute infrastructure and services supporting model inference, evaluation, and product delivery while collaborating with researchers and commercial teams to bring advanced molecular design tools to market.

What you'll do

Design and maintain compute and infrastructure systems for model inference and evaluation
Own end-to-end observability, alerting, and incident response for 24x7 production systems
Build developer-friendly services that scale from prototype to enterprise adoption
Translate scientific and product requirements into robust engineering solutions
Lead both greenfield infrastructure projects and scaling initiatives
Collaborate with product, research, and commercial teams on technical implementation

What they're looking for

Production systems architecture and deployment
Infrastructure observability and incident response
Cloud computing and containerization
AI/ML system scaling and optimization
Cross-functional communication and technical documentation
Security and platform engineering practices
Modern DevOps tooling and practices
On-call operations and reliability ownership

Benefits

Highly competitive compensation
Work on cutting-edge AI drug discovery technology
Mission-driven environment focused on therapeutic impact
Talented, fast-moving team with high ownership culture
Opportunity to shape infrastructure at scale
San Francisco office location

Likely interview questions

Walk us through a time you optimized model inference in production—what were the bottlenecks, and how did you approach latency vs. throughput trade-offs?
Describe your experience managing GPU infrastructure at scale. How have you handled GPU utilization, batching strategies, or cost optimization?

Tell us about a production incident you owned end-to-end. How did you debug it, what observability gaps did you find, and what did you change?
What's your experience with ML serving frameworks or inference engines? How would you evaluate one for a high-performance, multi-cloud deployment?
How do you approach building a system from scratch vs. scaling one that's already running? What's your mental model for knowing when to refactor vs. patch?
Describe a time you had to balance cost, latency, and reliability as competing constraints. How did you make the trade-off decision?
What's your experience with distributed systems failures—clock skew, partial failures, cascading outages? Give an example of one you've debugged.

Unlock all 7 questions free — and practice them live →

Apply on the employer's site →

Opens the official application on the employer’s site. No login required.