ABOUT ZAPDOS LABS

Zapdos Labs builds AI agents for factory and industrial operations. Our systems connect to existing cameras and operational data to help teams detect safety risks, understand production activity, investigate events, and act during the same shift.

We combine contemporary multimodal AI with precise computer vision. Our agents do more than generate text: they watch video, consult procedures, use software tools, take approved actions, and report outcomes with supporting evidence.

THE ROLE

We are looking for a product-minded Full Stack Engineer to build AI agents and the applications through which factory teams configure, supervise, and use them.

You will work across agent architecture, multimodal AI, backend services, integrations, and operator-facing applications. This is not primarily a dashboard role. You will build operational bots that can watch, reason, act, and report in real industrial environments.

WHAT YOU WILL BUILD

- Safety Officer Agents that detect hazards, reference the relevant SOP, save video evidence, alert supervisors, and track corrective action
- Shift Supervisor Agents that monitor production, investigate interruptions, explain likely causes, and prepare shift handovers
- Quality Inspector Agents that review production video, classify defects, track rejection rates, and identify process drift
- Maintenance Agents that observe equipment state, investigate abnormal behavior, recommend actions, and create work orders
- Inventory Agents that read labels, track movement, identify discrepancies, and maintain audit-ready records
- Factory Assistants that answer operational questions using live events, historical video, SOPs, manuals, and production records

WHAT YOU WILL DO

- Build multimodal agents that reason over video, images, documents, databases, and operational events
- Design agent tools for searching video, querying logs, reading SOPs, creating reports, and sending alerts
- Develop planning and orchestration flows for multi-step factory tasks
- Build operator-facing chat, review, approval,



and monitoring applications
- Show users what an agent observed, why it reached a conclusion, and what action it took
- Connect agents to cameras, computer vision models, customer documents, messaging systems, and factory software
- Implement human-in-the-loop controls for sensitive or uncertain decisions
- Build memory and context systems that preserve information across cameras, events, shifts, and facilities
- Develop evaluations for reasoning quality, tool selection, task completion, latency, and reliability
- Create guardrails that prevent unsupported conclusions or unsafe automated actions
- Analyze failures and improve prompts, tools, models, retrieval, data, and application logic
- Turn customer workflows into reusable agent capabilities instead of one-off demos
- Deploy and monitor agents operating continuously in real customer environments

WHAT WE ARE LOOKING FOR

- Strong full-stack software engineering experience
- Experience building production applications with LLMs, multimodal models, or agent frameworks
- Ability to design reliable APIs, backend services, databases, and user interfaces
- Understanding of tool calling, structured outputs, retrieval, memory, and agent orchestration
- Experience converting loosely defined user workflows into dependable software
- Strong product judgment, especially around when an agent should act, ask, or escalate
- Ability to debug failures across models, tools, data, prompts, and application code
- Comfort working directly with customers and learning unfamiliar operational processes
- Clear communication and reliable execution in an early-stage remote team




- Careful handling of confidential customer video and operational information

NICE TO HAVE

- Experience with TypeScript, React, Python, PostgreSQL, or comparable technologies
- Experience with vision-language models, video understanding, or computer vision
- Familiarity with MCP, tool-calling APIs, workflow engines, or agent evaluation frameworks
- Experience building copilots, autonomous workflows, or human-approval systems
- Familiarity with streaming video, asynchronous jobs, and event-driven architectures
- Experience integrating ERP, MES, CMMS, WMS, ticketing, or notification systems
- Knowledge of manufacturing, logistics, workplace safety, or industrial maintenance
- Experience deploying AI applications to edge devices or private customer environments
- Familiarity with authentication, role-based access control, audit logs, and enterprise security

WHAT SUCCESS LOOKS LIKE

During your first months, you will:
- Ship an agent that completes a real factory workflow from observation to action
- Connect the agent to video, operational documents, and at least one external tool
- Build a clear interface for reviewing its evidence, reasoning, and actions
- Establish repeatable evaluations for agent accuracy and task completion
- Reduce the manual configuration needed when deploying to a new facility

COMPENSATION AND STAGE

Zapdos Labs is an early-stage company building production AI systems for industrial customers. Compensation includes cash based on experience and commitment, meaningful equity reflecting early technical ownership, flexible remote work, and direct influence over product and architecture.

WHY THIS ROLE

Most AI agents operate inside browsers and documents. Zapdos agents operate at the intersection of software and the physical world.

You will build systems that observe real operations, reason through messy situations, and help factory teams make decisions during the same shift. The goal is not another chatbot. It is an operational teammate that factory personnel can understand and trust.

📌 Full Stack Engineer (India)
🏢 Zapdos Labs
📍 India