Tool & API Integration : Build and maintain the tools (function calling) that agents use to interact with our CRM, support ticketing systems, and internal databases.
RAG Pipeline Engineering : Develop and optimize Retrieval-Augmented Generation (RAG) pipelines to ensure agents have real-time, accurate context from our knowledge bases.
Connectivity & Orchestration : Implement the middleware that connects Agentic workflows to front-end support interfaces (Chat, Email, etc.).
Data Ingestion & Vectorization : Manage the lifecycle of data within our Vector Databases, ensuring high-quality embedding and retrieval performance.
Monitoring & Latency Optimization : Implement observability for AI calls (tracking tokens, costs, and response times) to ensure the super employee is as fast as a human, or faster.
Deployment & CI/CD : Manage the deployment of agentic microservices, ensuring that AI updates don't break existing support workflows.
Qualifications
API Mastery : Expert knowledge of RESTful APIs, Webhooks, and secure authentication protocols (OAuth, etc.).
The AI Stack : Hands-on experience with Vector DBs (Pinecone, Milvus, or Qdrant) and LLM providers (OpenAI, Anthropic, or local models).
Programming : Advanced Python (FastAPI, Pydantic) and experience with streaming data/WebSockets.
Framework Experience : Practical experience with LangGraph, Moltbot, or similar tool-calling frameworks.
AIOps : Familiarity with LLMOps tools for monitoring model performance and drift in production.
Additional Qualifications
Cognitive Architecture Design : Ability to design Multi-Agent Orchestration (MAO) patterns (e.g., Manager-Worker, Peer-to-Peer, or Hierarchical teams).
Advanced Prompt Engineering & Optimization : Mastery of DSPy (Programming instead of Prompting), chain-of-thought, and automatic prompt optimization.
Guardrail & Safety Engineering : Implementing frameworks like NeMo Guardrails or LlamaGuard to ensure agents don't hallucinate or leak sensitive trading data.
Evaluation (Eval) Frameworks : Building custom Eval suites using Ragas or TruLens to mathematically measure the accuracy and reliability of agent reasoning.
State Management : Expertise in managing long-term memory and persistent state across complex, multi-day agentic tasks.
Infrastructure : Experience with Docker, Kubernetes, and cloud-native serverless functions.
