Data scientist - Agentic AI
$155,500–$155,500 year
On-siteSan Jose, California, United States
Job Summary
Design and operationalize core agentic workflows for Marvis, Juniper's next-generation AI assistant, using LangGraph to enable autonomous network diagnostics and resolution. Develop and maintain MCP servers and skills, integrating LLM capabilities at production scale with robust error handling and retrieval services. Build evaluation frameworks for non-deterministic systems and collaborate with domain experts to translate troubleshooting playbooks into executable skills. Deploy containerized services in Kubernetes, manage CI/CD pipelines, and own observability for agent reliability. This role requires a Master's or PhD in a quantitative discipline with 4–6 years of production ML/AI experience, supported by an annual salary of $155,500–$315,000 in California.
Required Qualifications
- Master's or PhD degree in computer science, data science, mathematics, statistics, or a closely related quantitative discipline
- 4–6 years of experience building production ML/AI systems
- 1–2 years of hands-on work with generative AI and LLM-based applications
- Production experience with agentic orchestration frameworks: LangGraph, LangChain, Claude Agent SDK, or equivalent
- Solid understanding of agentic design patterns: ReACT loops, tool/function calling, dynamic tool binding, skill-based execution, multi-step planning, and self-correction
- Hands-on experience with MCP (Model Context Protocol) or equivalent tool-serving protocols: tool schema design, server implementation, registry management
- LLM API integration at scale: prompt engineering, structured outputs, streaming, error handling, and cost optimization
- RAG pipeline design: chunking strategies, re-ranking, hybrid search, vector stores (OpenSearch or equivalent), and relevance optimization
- Experience building evaluation and testing frameworks for non-deterministic AI systems (offline evals, A/B testing, LLM-as-judge)
- Strong foundation in statistical and machine learning techniques — anomaly detection, time-series analysis, clustering, causal inference, or related methods
- Applied ML intuition: knowing when to use retrieval vs. fine-tuning, prompt engineering vs. structured generation, and how to debug model behavior in production
- Proficient Python developer with experience in production codebases (not just notebooks)
- Kubernetes: deploying, scaling, and managing workloads (Deployments, Services, ConfigMaps, Secrets, health probes)
- CI/CD pipelines for automated build, test, and deploy (Jenkins, GitHub Actions, ArgoCD, or similar)
- Container image management: building, tagging, versioning via Docker; familiarity with a container registry (ECR, GCR)
- Backend service development: FastAPI or equivalent; REST/GraphQL API design
- Observability for AI systems: experience with tracing, monitoring, and logging tools (LangFuse, Prometheus, or equivalent)
Desired Qualifications
- LangGraph (strongly preferred)
- Experience with agent memory systems (e.g., LangMem, custom memory architectures)
- Familiarity with sandboxed code execution environments (E2B, Firecracker, or similar)
- Networking domain knowledge (wireless/wired diagnostics, network troubleshooting)
- Experience with AWS Bedrock, OpenSearch Serverless, or similar managed AI/ML services
- Great written and verbal communication skills; ability to articulate technical designs to senior leadership
Hiring someone like this?
Get your role in front of qualified candidates on Sorce.