Hybrid Agentic Analytics Engineer: AI Agents

Contract Length: 1 year with probability of extensions Location: Onsite in South San Francisco, hybrid Hours: 9-5 PST Start Date: ASAP Client Overview: As an Agentic Analytics Engineer in Synthetic Molecule Pharmaceutical Sciences (SMPS), you will be responsible for integrating scientific and business data from multiple sources to generate agentic analytics product. You will be connecting Large Language Models (LLMs) to enterprise tools, databases, and APIs to enable conversational data analytics, and contributing to the development of life-saving therapeutics. Job Description: Agent Orchestration: Use frameworks like LangChain , CrewAI , or LangGraph to build specialized agents Tool Integration: Implement the Model Context Protocol (MCP) to connect agents to internal systems like Microsoft Suite, Slack, or SQL databases. Workflow Mapping: Translate complex business processes (like "genotoxic impurity profile of a therapeutic molecule") into a series of logical steps an AI can follow. RAG Pipeline Maintenance: Build and optimize Retrieval-Augmented to give agents access to enterprise knowledge and data. Evaluation & Debugging: Use "LLM-as-a-Judge" frameworks to test agent reliability, find where reasoning loops fail, and implement "Human-in-the-Loop" (HITL) approval gates. Skill Set Requirements: Language: Python and SQL (advanced) Data Stack: Modern data warehouses such as Snowflake Frameworks: LangChain and LangGraph Orchestration: Mastery of reasoning loops (ReAct, Chain-of-Thought) Protocols: Model Context Protocol and RESTful API integration Nice to Haves: Basic understanding of chemistry and drug development and relevant experience in biotech or pharmaceutical companies Education/Certification(s)/License(s): Education: B.S./M.S./Ph.D. in Computer Science, Data Science, or a related field, with 1+ years of professional work experience Portfolio: Evidence of building at least one autonomous system (e.g., an agent that can successfully scrape a site, summarize findings, and email a report #J-18808-Ljbffr Medasource