Note: The job is a remote job and is open to candidates in USA. BullFrog AI develops machine-learning platforms for drug discovery and clinical decision support. They are seeking a Staff Platform Engineer to lead the evolution of their decision-adjudication platform, bfARENAS, focusing on algorithm design, infrastructure, and user interfaces for pharmaceutical clients.
Responsibilities
- Advance the core algorithms by improving ranking quality, uncertainty quantification, and scaling characteristics; design and run evaluations that distinguish genuine algorithmic improvements from prompt-level artifacts
- Build the agent layer by developing and refining the LLM-powered components that structure decision problems before adjudication and that maintain quality during execution
- Construct human-in-the-loop interfaces that let domain experts contribute pairwise judgments, review and override automated decisions, and configure problem structure for new datasets
- Own the platform infrastructure including multi-provider LLM orchestration, structured-output handling, run reproducibility, cost tracking, telemetry, and resilient execution
- Deliver client-facing outputs such as interactive visualizations, drill-down dossiers, and uncertainty-aware reporting that pharma scientists trust enough to make decisions on
- Maintain a standing quality discipline through bias detection, redundancy analysis, judgment diagnostics, and multi-judge reliability — built into the platform rather than performed ad hoc
Skills
- 6+ years of production software engineering experience, with at least 2+ years building LLM-powered systems that other people depend on — not prototypes, not demos
- Strong Python coding experience. The codebase is Python end-to-end
- Real statistical literacy. Comfort with ranking and pairwise comparison models, bootstrap methods, and the difference between a misleading confidence interval and a useful one. Computer-science-with-statistics or statistics-with-computer-science backgrounds are welcome
- Demonstrated ability to design evaluation harnesses for LLM systems. 'I designed the eval that proved the prompt change was real' rather than 'I tried some prompts.'
- Product instinct. You will be building interfaces that determine whether scientists trust the system enough to act on its outputs
- Comfort owning a system end-to-end across algorithms, infrastructure, and user-facing experience. This role is not split into three
- Domain exposure to drug discovery, clinical development, or computational biology. Not required but shortens the ramp
- Background in ranking systems, multi-criteria decision analysis, social choice theory, recommender systems, or tournament/matchmaking design
- Experience shipping client-facing analytical products in regulated or scientific settings
- Familiarity with multi-provider LLM orchestration (OpenAI, Anthropic) and structured-output workflows
- Prior published work: engineering, research, or both
Benefits
- 15 days of paid time off annually
- 11 paid holidays annually
- Medical, Dental, and vision coverage with eligibility on the first day of employment
- Short-Term Disability
- 401 (k) with enrollment upon day one
- Eligibility for bonus and stock options based on a combination of individual and company performance.
Company Overview