AI/ML Engineer

Company Overview Codvo is a global empathy-led technology services company where software engineering excellence and human-centered innovation come together. Our mission is to accelerate our clients’ digital transformation through world-class design, cloud engineering, data modernization, digital engineering, and enterprise AI. As we deepen our focus on the Oil & Gas sector, we partner with upstream, midstream, downstream, LNG, chemical, and pipeline operators to unlock measurable value across their digital and AI transformation journeys. Role Summary You will work day to day with the Baker Hughes lead AI engineer, building and operating agentic AI workloads in production. The job is hands-on: agent workflows, retrieval pipelines, prompt and context engineering, connectors, and the operational work that keeps LLM applications healthy in production. Responsibilities • Build and maintain multi-agent workflows on LangGraph, including tool definitions, state handling, and error recovery. • Minimum 2 yrs of experience in Agentice AI and Proven projects on Agentic AI. • Build retrieval pipelines: document chunking, embeddings, vector search, and retrieval quality evaluation. • Apply context engineering practices so each query sends only the minimum required context to the model, keeping token cost low. • Implement caching and curated-answer paths so repeated or well-known queries avoid redundant LLM calls. • Build connectors from agent applications to enterprise systems and customer data sources. • Instrument agents with Langfuse and Arize traces; triage production issues from those traces. • Write evaluation suites for agent behavior and run them before every release. • Expose agent capabilities through FastAPI services with clean request and response contracts. Must-Have Skills and Experience • 5 years of software engineering with at least 1 year building LLM applications. • Strong Python and Git discipline; comfortable in code review. • Hands-on LangChain or LangGraph experience (or an equivalent orchestration library with a clear story of what you built). • Vector database experience (pgvector, Qdrant, Weaviate, or similar) and a working grasp of embedding models. • REST API development, ideally FastAPI. • SQL basics and comfort reading unfamiliar schemas. Nice to Have • vLLM or other self-hosted inference serving. • Redis for caching or queues. • Kubernetes or OpenShift deployment experience. • LLM evaluation tooling (Langfuse evals, Ragas, promptfoo, or similar). • Exposure to Databricks or lakehouse data sources.