AI-ML Engineer

Role Overview

We're hiring an AI/ML Engineer to build intelligent, agent-driven capabilities into Apptimus spanning ML models for prediction/anomaly detection and agentic workflows for autonomous reasoning and task execution across our IoT and facility data.

Responsibilities

• Design and build agentic AI workflows (LangChain/LangGraph/CrewAI) for multi-step reasoning, tool-calling, and automated decision-making on Apptimus data
• Develop and deploy ML models (forecasting, anomaly detection, classification) on IoT sensor and facility usage data, and wire them into agent pipelines as callable tools
• Write production Python for agent orchestration, model serving, and backend/API integration
• Write SQL for data extraction, feature engineering, RAG grounding, and reporting
• Implement prompt engineering, memory management, and guardrails to keep agents reliable in production
• Build and maintain embeddings/vector search pipelines to ground agents in Apptimus data (occupancy, energy, checklist/license usage, etc.)
• Evaluate agent and model performance (accuracy, latency, cost) and iterate through structured testing

Requirements

• 2+ years in AI/ML or software engineering, with hands-on exposure to LLM-based or agentic systems
• Strong Python — ML libraries (scikit-learn, pandas, PyTorch/TensorFlow) plus LLM SDKs (OpenAI, Anthropic)
• Good SQL — complex queries, joins, relational DBs
• Solid understanding of LLM concepts: prompting, embeddings, vector DBs, RAG
• Practical experience with at least one agent orchestration framework (LangChain, LangGraph, CrewAI, AutoGen)
• Fundamentals of ML model development and evaluation (regression/classification, time-series, or anomaly detection)

Nice to Have

• Hands-on experience shipping a production-ready agentic model/system — i.e., an agent that went beyond a prototype into a live, monitored deployment handling real users/data
• IoT/smart building or sensor data experience
• MCP or similar tool-calling protocols
• Docker, CI/CD, LLM observability tools (LangSmith, Langfuse)
• Experience combining classical ML models with LLM agents (model-as-tool patterns)