SDE 3

We are seeking an SDE 3 (Agentic Orchestration and Distributed Systems) to build and scale the core orchestration engine that powers our multi-agent AI.
procurement platform. In this role, you will design robust, stateful execution graphs and durable workflow backends that allow autonomous agents to plan, reason, negotiate, and execute complex enterprise procurement workflows from supplier discovery and dynamic RFx generation to contract compliance checks and ERP data synchronization.
This role sits at the intersection of modern production Python and enterprise-grade distributed systems, requiring an engineer who treats agentic logic with the rigor of high-reliability backend engineering.
Responsibilities:
Agent Execution and Graph Orchestration: Design, implement, and optimize stateful multi-agent systems (using frameworks such as LangGraph, AutoGen, LlamaIndex, or custom DAG runtimes) capable of handling long-running, non-deterministic tasks with dynamic planning and tool-calling loops.
Durable Execution and State Management: Build distributed, fault-tolerant workflow pipelines leveraging event brokers and durable execution engines (e. g., Temporal, Celery/Redis, Kafka) to manage state persistence, timeouts, retries, and checkpointing.
Human-in-the-Loop (HITL) Architectures: Architect clean state pause/resume, human approval gates, and escalation mechanics to ensure safe, auditable human intervention during autonomous workflows.
Deterministic Tooling and Enterprise Integration: Develop strongly typed, schema-validated APIs (via Pydantic, Instructor, and structured outputs) to interface agents with enterprise backends, ERP systems (SAP, Coupa, and Workday), and external vendors.
Observability, Evaluations, and Guardrails: Implement deep distributed tracing (OpenTelemetry, Langfuse), automated LLM evaluation harnesses, dynamic fallback models, and token/rate-limit governance to ensure low latency and high reliability.
Technical Leadership: Lead design reviews, define coding standards for asynchronous Python at scale, and mentor SDE 1/2 engineers across backend and AI engineering domains.
Requirements:
Experience: 8+ years of core software engineering experience building distributed backend systems, with recent hands-on experience designing production LLM agents or complex asynchronous workflow engines.
Deep Python Expertise: Expert-level mastery of modern Python, including asyncio, concurrency patterns, type annotations (mypy), Pydantic v2 and high-performance frameworks like FastAPI.
Distributed Systems and Storage: Proven track record with distributed caching, relational/document databases (PostgreSQL, Redis), and message queues/event streaming (Kafka, RabbitMQ).
Durable Workflow Familiarity: Hands-on experience or deep architectural understanding of durable execution frameworks (e. g., Temporal, Cadence, Airflow) or persistent state-machine patterns.
AI and Agent Ecosystem: Deep familiarity with agentic paradigms (ReAct, multi-agent coordination, reflection), vector search/RAG pipelines, and LLM evaluation suites.
Engineering Standards: Passion for deterministic validation, comprehensive test coverage (unit/integration/evals), CI/CD automation, and low-latency API design.
Preferred Qualifications:
Polyglot exposure to secondary languages used in high-throughput enterprise infrastructure (Go, TypeScript, or Java).
Prior domain experience in enterprise B2B SaaS, supply chain orchestration, or procurement technology.

Skills
Agentic AI, Python

Experience
8-12 yrs