AI Engineer (Supply Demand Management)

the foundational engineer behind our AI Tool Foundry-a centralized platform of Python-based functions, APIs, and wrappers that our intelligent agents use to fetch data, interact with external systems, and execute complex workflows. You will design how tools are standardized, how they are described to LLMs (via schemas and docstrings), and how we ensure agents execute them safely and predictably.
If you love writing pristine, highly typed Python code and are fascinated by the intersection of traditional backend engineering and modern LLM function calling, this is the role for you.","responsibilities":"Build the AI Tool Foundry: Architect and maintain a centralized repository of modular, reusable Python tools. Establish the standards, base classes, and registries that allow any internal AI agent to easily discover and consume these capabilities.

Develop Agent-Facing APIs & Functions: Write robust Python functions and external API wrappers specifically designed for LLM consumption. You will obsess over developments of tools to ensure LLMs rarely hallucinate arguments.

Safety, Sandboxing & Permissions: Design execution environments that allow agents to act autonomously while maintaining strict guardrails. Implement human-in-the-loop (HITL) checkpoints, rate limits, and scope-based permissions for agentic actions.

Testing & Evaluation: Traditional CI/CD isn't enough for non-deterministic agents. You will build testing frameworks that not only unit-test the Python functions but also evaluate how reliably the LLM selects the correct tool and provides the correct parameters.

Observability & Tracing: Implement logging and tracing (e.g., via LangSmith, Phoenix, or OpenTelemetry) to monitor agent tool usage, track latency, catch parsing errors, and debug multi-step agent reasoning loops.

Preferred Qualifications
Experience with agentic frameworks (e.g., LangChain, LlamaIndex, AutoGen, CrewAI) and a strong opinion on when to use them versus when to write custom orchestration.

Familiarity with AST (Abstract Syntax Trees) or dynamic code execution/sandboxing (e.g., executing agent-generated Python code safely).

Experience building retrieval systems (RAG) or working with Vector Databases (Pinecone, Qdrant, Weaviate).

Minimum Qualifications
Expert-Level Python: 4+ years of professional software engineering experience, with deep expertise in Python. You should be highly proficient with modern Python paradigms, including asyncio, type hinting, decorators, and advanced OOP/functional patterns.

Pydantic & Data Validation: Extensive experience using validation libraries to strictly define data models, parse inputs, and handle serialization/deserialization.

LLM Function Calling Experience: Hands-on experience working with the function-calling or tool-use capabilities of modern LLMs (e.g., Gemini's tools API, or open-source equivalents).
API & Systems Integration: Deep experience integrating with third-party APIs (REST, GraphQL, gRPC) and handling edge cases like pagination, backoffs, and chaotic external state.

Backend Best Practices: Strong fundamentals in API design (FastAPI/Starlette), version control, CI/CD, and writing comprehensive test suites (Pytest).","internalDetails":null