Building AI dev tools and agentic systems—hands-on with LLMs, memory, tool integrations, and tracing/observability.
About the Role
Build and maintain agentic LLM-powered systems that assist proposal writing for the AEC industry. The role focuses on implementing modular agents, memory layers, tool integrations, observability, and production-grade quality for multi-agent orchestration and retrieval-backed workflows.
Job Description
Role
Kantiv is hiring an Agentic Systems Engineer with ~2–4 years of experience to help build the next generation of agentic applications for proposal writing in the AEC industry. The work centers on multi-agent orchestration, memory and retrieval layers, tool integrations, and the production plumbing that keeps agentic systems reliable.
Key Responsibilities
- Build modular, plug-and-play agents that integrate into the broader stack.
- Add and manage memory layers (short-term, long-term, summarization, retrieval-backed).
- Wire up tool integrations, MCP servers, and agent skills.
- Ensure feature quality via tests, evals, observability, and monitoring.
- Investigate production traces, debug issues, and implement fixes.
Requirements
- 2–4 years writing production software.
- Strong Python skills; ability to write idiomatic, well-structured, testable Python.
- Solid grounding in agentic and LLM concepts: RAG, prompting patterns, tool use, structured outputs, streaming, and context management.
- Experience building with modern agent toolkits (side projects, prototypes, or production work).
- Ability to move quickly through unfamiliar codebases and pragmatic problem solving.
- Data-driven approach: comfortable using production traces, eval numbers, and logs to direct decisions.
- Hands-on experience with LLM tracing/observability tools such as Langfuse or LangSmith.
Preferred / Nice-to-have
- Experience with search and retrieval systems: embeddings, vector databases, hybrid retrieval, rerankers.
- End-to-end experience designing and running LLM evaluations (eval design, harnesses, measurement strategies).
- Deep experience with LangGraph: custom graphs, checkpointers, context-management nodes, and state pruning.
Interview Process
- 30-minute introductory Zoom
- 45-minute Python / agentic coding proficiency test (2 problems)
- 60-minute deep-dive project presentation (20–25 minute presentation + Q&A)
- 45-minute Gen AI / LLM fundamentals interview
- 30-minute culture fit interview