Software Engineer, Agents

About Mercor

Mercor aims to organize human intelligence as a foundational element powering the AI economy. The company operates a platform where millions of domain specialists generate over $4 million daily by training leading AI models. Through the APEX benchmark family, Mercor quantifies AI's practical effects on professional tasks, and its Mercor Enterprise segment aids Fortune 500 clients in capturing top talent expertise and converting it into AI agents.

Mercor is pioneering a fresh work model where expertise directly advances AI, requiring a dedicated, fast-moving team collaborating with researchers, operators, and AI companies shaping transformative systems globally. Mercor, a profitable Series C startup valued at $10 billion, has offices in San Francisco, New York City, and London supporting in-person work five days a week.

Role Overview

We seek an experienced engineer to develop scalable agent-oriented products. Your work will span backend technologies such as Python, FastAPI, Django, Pydantic; frontend frameworks including Next.js, React, TypeScript, Tailwind; databases like PostgreSQL, MySQL, Snowflake, DuckDB, Redis; and orchestration/infra tools such as Kubernetes, Temporal, Modal, Woz. You'll also leverage agent and LLM platforms LangGraph, LangChain, FastMCP, Harbor, NemoGym alongside observability tools Datadog, PostHog, and LangSmith.

Post-hiring, you'll be assigned to the team best suited for your impact:

• Automation team: develops intelligent automation solutions for operational activities, emphasizing business results and cross-departmental collaboration, focusing on 0-to-1 product creation.
• Studio team: manages evaluation and annotation platforms for reinforcement learning environments and tasks, partnering with frontier AI researchers to enhance next-gen models.

Responsibilities

• Fully manage feature development of agent-based systems from concept through implementation, launch, and feedback-driven improvement.
• Design and deploy LLM agents, agent harnesses, and verifiers, including tools, prompts, and governance policies to ensure dependability.
• Implement Python/FastAPI services and Temporal/Modal pipelines orchestrating agent execution and human-in-the-loop processes.
• Create state-of-the-art RL environments that simulate enterprise applications, virtual colleagues, and data-rich contexts to support extended tasks.
• Develop tooling that transforms agent behavioral data into actionable insights via statistical analysis and automated failure detection.
• Build comprehensive full-stack interfaces using Next.js/React facilitating operator-agent collaboration, and evolve data structures supporting agent context and auditability.
• Establish agent quality metrics and foster ongoing enhancements by constructing evaluations, instrumenting traces, identifying failure modes, and refining prompts and safeguards to optimize reliability, cost, latency, and user experience.
• Collaborate cross-functionally with Product, Design, Research, and Operations teams to define agent autonomy boundaries, balancing automated actions with human oversight workflows.

Why Join Mercor

• Drive meaningful impact in the training and testing of leading AI innovations worldwide.
• Gain early access to cutting-edge model capabilities months ahead of the market.
• Engage in both infrastructure and research-related projects with rapid advancement opportunities.

Benefits

• Performance-based biannual bonuses
• Generous equity grants vesting over four years
• Relocation bonus up to $15,000
• Housing stipend of $10,000 if residing within half a mile of office
• $1,500 monthly meals stipend
• Complimentary Equinox gym membership
• $200 monthly laundry reimbursement
• $200 monthly personal wellness reimbursement
• Comprehensive health, dental, and vision insurance