Marcus Intelligence

About Marcus Intelligence: Marcus Intelligence builds AI-native platforms for real-world engineering. We’re now on a mission to push the boundaries of agentic automation—combining distributed LLMs, vision, and reinforcement learning into end-to-end autonomous QA and development systems. Role Description: We need someone who has already built, led, or shipped multi-agent systems at scale , ideally with modern frameworks (LangGraph, LangChain, CrewAI, OpenAgents) and real knowledge of the tradeoffs in agent architecture. What You’ll Own: Architect, implement, and scale multi-agent workflows for autonomous software QA/testing Build and optimize LangGraph-based orchestrators (or similar), dynamically managing LLMs, utility agents, annotators, and feedback loops Drive research→deployment: rapidly prototype and ship agentic RL/self-healing, tool-calling, and hybrid-memory integrations Mentor and code-review junior AI/infra teammates, guiding best practices for parallelism, persistence, security, and traceability Integrate agent pipelines with CI/CD, vector stores, and log/monitoring stacks (e.g., Redis, Pinecone, OpenTelemetry) Must-Have Skills: deep experience shipping agentic LLM/AI workflows (not tutorials; real production) Expert with LangGraph/LangChain (incl. custom graph topologies, step handlers, streaming, persistent memory) Solidity with async orchestration, task queues, distributed systems (FastAPI, Celery, or event-native stacks) Hands-on with cross-agent state sharing, error isolation, and message routing Solid Python (>=3.10), with strong design and debugging habits Ability to self-manage, engineer under ambiguity, and document for others Good to Have: Real vision for long-running stateful agents (agents that persist across sessions/runs) Prior work on agentic RL/self-healing, or at least background in agentic PEFT techniques Exposure to advanced tool calling, sandboxing, or mechanical feedback (memory, retrievers) OSS contributions in agentic LLM frameworks (link portfolio/GitHub) Offer: Be the core designer and builder of Marcus Intelligence’s next-gen agent layer Freedom to experiment, build, and shape system architecture from MVP to scale Flexible remote/contractor options (if working from India timezone or willing to overlap) If desired, opportunity to transition to founding engineer, with equity and technical leadership Compensation: First 2 months may be unpaid/trial (if part-time or contract), with clear paid/equity pathway after mutual fit/proven impact Will pay top-of-market for the right person, including incentives for OSS/community value