About the ClientOur client is an AI and automation company focused on solving complex problems in heavy industries such as utilities, oil & gas, logistics, and manufacturing. The company builds end-to-end AI platforms, custom models, and enterprise AI strategies that help organizations optimize operations, reduce downtime, and improve efficiency. Its solutions include predictive maintenance, supply chain optimization, fleet management, and outage response systems powered by advanced data and machine learning. They aim to “rebuild the industrial base” by combining deep industry expertise with cutting-edge AI engineering to deliver measurable operational outcomes for large enterprises. They're building the first industrial OS and emerging from stealth. This isn't incremental - it's zero-to-one product development.What we’re looking for3+ years experience (5+ preferred) building and shipping production AI systems, ideally agentic productsExperience as an early engineer at a VC-backed startup, taking systems from 0 to 1Comfortable working autonomously in a fast-paced, high-intensity environment with high ownershipTrack record building enterprise AI platforms that handle complex, physical-world data (video, IoT, sensor streams)What you have doneBuilt and shipped agent systems in production with orchestration, tool use, state management and human-in-the-loop workflowsWorked with physical-world data at scale (RTSP video, time-series telemetry, vibration, thermal, PLC / OPC-UA) under real constraints (noise, drift, latency, alignment)Built multimodal AI pipelines combining vision (detection, segmentation, action recognition) with structured operational dataDesigned or contributed to ontology, knowledge graph or structured context systems grounded in real asset hierarchies and processesIntegrated AI systems with ERP, CMMS, WMS, historians or PLC layers and handled normalization and schema mappingShipped end-to-end systems from ingestion and model serving through backend services (Python, TypeScript) to frontend interfacesWhat you will buildAgent runtime and orchestration across Vision Quality, Predictive Maintenance and Operations Planning agentsContext assembly from ontology and memory, tool dispatch, approval gates, tracing and cost controlsMultimodal sensor pipelines: video processing, YOLO / segmentation, FFT feature extraction and cross-sensor correlationIndustrial ontology / knowledge graph mapping plants, assets, sensors, work orders, materials and maintenance historyMemory layer including trace storage, playbooks, asset templates and transferable failure pattern librariesOperational decision surfaces: dashboards, alerting workflows with evidence, replanning tools and audit trailsIntegration connectors across ERP, CMMS, WMS and PLC systems into a unified schemaWho you areHave built AI systems from 0 to production in real environmentsStrong background in AI/ML applied to physical-world systemsUnderstand that the hardest problems are in data ingestion, normalization, temporal alignment and ground truthThink in terms of production systems: reliability, failure modes, cost and human interactionComfortable operating in a high-intensity, fast-moving environment with significant ownershipMotivated to work on real-world industrial problems that require both depth and executionTech stackTypeScript, Python, FastAPI, Agentic AI, OpenCV, PostgreSQL, React NativeWhy candidates should joinCategory-defining opportunity: You'll help build the first industrial operating system - there's no real OS for industry yet and this will exist in manufacturing plants and chemical facilities over the next 10-20 years.Exceptional equity package: Equity pool for employees is double that of the industry standard, with $200k+ equity value from day one.Elite team with real industry credibility: Work alongside ex-Palantir engineers including a former VP who oversaw robotics at ABB across 90 markets. Team includes autonomous driving engineers from Audi/Volvo and researchers from Harvard/MIT.Real-world impact: Connect to incredible amounts of unused industrial sensors and make that data useful for heavy industry and manufacturing - solving problems that matter in the physical world.
