Applied AI Engineer

We are seeking an Applied AI Engineer to build, evaluate, and deploy production AI applications using large language models, AI agents, and Retrieval-Augmented Generation (RAG). The role requires strong software engineering skills and hands-on experience translating business requirements into reliable AI workflows. You will collaborate with product and engineering teams to deliver measurable business outcomes.
Responsibilities
Design and implement AI agent workflows using LangGraph, including nodes, edges, conditional routing, and state management.
Build RAG pipelines covering ingestion, chunking, embeddings, vector indexing, hybrid search, reranking, and answer generation.
Implement input validation, intent detection, query classification, query rewriting, history-aware retrieval, and semantic caching.
Integrate LLMs with APIs, databases, tools, and enterprise data sources.
Manage conversation history, agent memory, persisted state, and recovery from interrupted workflows.
Create evaluation datasets and measure retrieval quality, answer quality, and agent task completion.
Use Langfuse or equivalent tools to trace LLM calls, monitor latency and token costs, and investigate failures.
Diagnose production issues such as incorrect retrieval, hallucinations, tool failures, and state inconsistencies.
Handle API rate limits, timeouts, and transient failures through appropriate retries, backoff, concurrency controls, and fallbacks.
Collaborate with product and engineering teams to deliver measurable business outcomes.
Requirements
Proficiency in Python or TypeScript/JavaScript.
Experience building backend services, REST APIs, and asynchronous workflows.
Practical knowledge of testing, debugging, logging, Git, and deployment.
Ability to explain implementation choices and tradeoffs in previously delivered projects.
Hands-on experience building agent workflows with LangGraph.
Understanding of graph state, nodes, edges, conditional routing, and workflow termination.
Experience with tool calling, structured outputs, state persistence, and error recovery.
Understanding of the distinction between conversation history, agent memory, and caching.
Experience implementing and debugging end-to-end RAG pipelines.
Understanding of embeddings, vector dimensions, cosine similarity, and similarity search.
Familiarity with a vector database or search engine such as pgvector, Qdrant, Pinecone, Weaviate, or Elasticsearch.
Ability to select and justify chunking strategies, chunk sizes, and overlap.
Experience with metadata filtering, keyword and semantic retrieval, hybrid search, and reranking.
Ability to build representative evaluation datasets with expected answers or relevance labels.
Understanding of retrieval metrics such as Precision@K, Recall@K, MRR, and nDCG.
Experience evaluating answer correctness, relevance, groundedness, and citation accuracy.
Ability to evaluate agent task success, tool-selection accuracy, reliability, latency, and cost.
Familiarity with human review, LLM-based evaluation, and regression testing.
Experience with Langfuse or comparable tracing and monitoring tools.
Ability to trace failures across retrieval, prompts, model responses, and tool execution.
Understanding of rate-limit handling, exponential backoff with jitter, retry limits, and request concurrency.
Awareness of prompt injection, sensitive-data handling, and access control in AI applications.

Work arrangement
No