Cheiron
Gangnam-gu, South Korea · On-site · Full-time
Loading job details
Preparing the role information
Backend Engineer, Applied Agents Seoul, South Korea · Engineering · Full-time · Fully in-person Company Overview Cheiron is building an AI-native operating system for drug development: a connected view of the evidence, assumptions, risks, decisions and commitments behind a drug program. The goal is to help biopharma teams understand what is known, test their assumptions and see how a change affects the rest of a program. Our current focus is CMC (Chemistry, Manufacturing, and Controls): how a drug is made, tested and kept consistent throughout its commercial life. A manufacturing change can create a chain of evidence, review and submission work across markets. Today, teams piece this work together across documents, spreadsheets and institutional memory. We're building agents that connect the relevant evidence, reason through requirements and help teams prepare and review the resulting submissions. Founded in 2024 by Stanford-trained AI researchers, Cheiron is a Seed-stage company headquartered in Los Altos, California. Our team of 19 people, including 8 engineers, combines AI and software engineering with experience from pharma and biotech. The platform has been adopted by tens of thousands of biopharma professionals. With $16M total raised, including an $11M seed round led by Menlo Ventures, we're growing our engineering team in Los Altos and Seoul. About the Role Build the agent execution and backend systems that turn complex regulatory work into reliable, reviewable software workflows. Cheiron is building the governed execution layer for post-approval regulatory submissions, with its deepest reasoning in CMC change. You will own substantial parts of the product: how an agent decomposes work, retrieves evidence, calls tools, preserves state, recovers from failure and produces a result a regulatory professional can inspect. The work spans agent runtimes, backend services, retrieval, data pipelines and production operations. You will work directly with founders, engineers, domain experts and customers to turn ambiguous needs into useful capabilities. This is broad early-team responsibility with founding-engineer ownership. We value people who understand unfamiliar systems quickly, reason carefully about how they work, and carry that reasoning through to tested software. You will use AI-assisted development actively while remaining accountable for the design, code and results. What You'll Build Agent execution: runtimes and harnesses for multi-step work, tool calling, isolated code execution, context management, retries and recovery. Backend foundations: services, APIs and data models using Python, FastAPI and Postgres, with clear contracts, durable state and useful observability. Evidence and retrieval: ingestion, versioning, indexing and search over regulatory and life-sciences sources; evaluate retrieval quality, latency and failure cases. Traceable decisions: link assessments and generated content to source evidence, retain the relevant versions and uncertainty, and make outputs inspectable by domain experts. Production reliability: regression tests and evaluations, failure diagnosis, monitoring and improvements that make real customer workflows dependable. Reusable product capabilities: translate customer needs into the shared product, prioritize a useful scope, ship it and learn from its operation. What We're Looking For Software depth: 2+ years shipping and operating substantial production software. You can explain your own implementation, technical decisions and their effects, from the request or input through state changes and outputs. Strong fundamentals: sound programming, API design, data modeling, databases, error handling, debugging, testing and system reliability. Fast learning: quickly understand an unfamiliar system, build an accurate working model and apply new concepts to working code. Deep reasoning under uncertainty: distinguish observations from assumptions, form and test explanations, notice consequential unknowns and revise your approach when evidence changes. Ownership and judgment: make progress with incomplete specifications, choose sensible trade-offs, communicate limitations, seek help when needed and follow through to a usable result. AI-assisted engineering: adopt AI-assisted software workflows quickly, direct tools deliberately, inspect generated changes and verify behavior against the actual system. Clear collaboration: explain technical choices, give useful feedback on code and work closely with people who bring different technical and domain expertise. Currently resident in the country of the selected job: the United States for Los Altos or South Korea for Seoul. The role is fully in person; relocation support covers moves within that country. Working English sufficient to discuss technical designs, write documentation, and collaborate with teammates across the US and Korea Useful Experience Production agents, workflow orchestration, tool execution, retrieval, evaluation or
Cheiron
Posted via Ashby_ats
Apply Now opens Rozgoo's sign-up and application flow for this role. Updated yesterday.
Apply Now