Principal AI Engineer

eino.ai · New York, NY · 3 wk ago

EngineeringFull-time

About The Role

This role owns the design and implementation of AI agents that power intelligent workflows across our platform, from reasoning and automation to evidence-driven RCA, customer-facing copilots, internal operations, and product intelligence. You’re a strong Python engineer who understands modern agentic AI architecture and can turn ambiguous product needs into production-grade systems. You are comfortable building agents that use tools, retrieve context, reason over structured and unstructured data, execute workflows, and integrate deeply with backend services. This is not a prompt-only role. We’re looking for someone who can architect, build, test, deploy, and operate AI-powered features end to end.

What You’ll Build

Agentic AI systems
- Production-grade AI agents for different product and internal applications:
  - Site-level connectivity analysis
  - Evidence-first root cause analysis workflows
  - Customer-facing assistant and copilot experiences
  - Internal automation for operations, support, and engineering workflows
  - Reasoning workflows over telemetry, incidents, site data, and knowledge sources
- Modern agentic architecture:
  - Tool/function calling
  - Planning and execution loops
  - State and memory management
  - Retrieval-augmented generation
  - Structured outputs
  - Workflow orchestration
  - Multi-agent patterns where appropriate
  - Guardrails, permissions, and safety constraints
  - Observability and evaluation frameworks
- AI-native product primitives:
  - Agent task models
  - Context assembly pipelines
  - Tool registries
  - Prompt/version management
  - Human-in-the-loop review flows
  - Agent execution traces
  - Eval datasets and regression testing
Python backend + product engineering
- Backend services and APIs in Python:
  - FastAPI/Django/Flask-style services
  - Workers and async jobs
  - Integrations with internal systems and external APIs
  - Data models supporting AI workflows
  - Event-driven and workflow-driven architectures
- Retrieval and knowledge systems:
  - Vector search
  - Hybrid search
  - Document ingestion
  - Chunking and indexing strategies
  - Metadata filtering
  - Grounding and citation workflows
- Production AI infrastructure:
  - LLM provider integration
  - Model routing
  - Cost and latency optimization
  - Caching
  - Rate limits and retries
  - Monitoring and debugging
  - Failure handling and fallback behavior
- Feature delivery end to end:
  - Product scoping
  - Architecture
  - Implementation
  - Testing
  - Deployment
  - Observability
  - Iteration based on user feedback

Responsibilities

Architect and implement production-grade AI agents that solve real business and product problems.
Build agent workflows that can reason over Eino’s data, tools, telemetry, site models, incidents, and knowledge sources.
Own agent architecture patterns across planning, memory, retrieval, tool execution, structured outputs, evals, and observability.
Build and operate core Python backend services that support AI-powered product features.
Work closely with product, engineering, and leadership to identify high-value agentic AI use cases.
Move quickly from prototype to production while maintaining reliability, security, and maintainability.
Establish testing and evaluation discipline for AI systems:
- Unit/integration tests
- Prompt and workflow regression tests
- Agent evals
- Golden datasets
- Trace review
- Failure analysis
Drive practical AI engineering standards:
- Correctness over demos
- Grounded outputs
- Measurable quality
- Clear contracts between agents, tools, and backend services

Required Qualifications

Strong experience building production Python systems, including services, APIs, workers, and backend infrastructure.
Hands-on experience building AI agents, LLM-powered applications, RAG systems, workflow automation, or tool-using AI systems.
Deep familiarity with modern agentic AI architectures, including tool/function calling, planning and execution loops, state and memory management, retrieval, structured outputs, guardrails, observability, and evaluation frameworks.
Ability to take features from ambiguous requirements to production deployment.
Strong software engineering fundamentals: system design, testing, debugging, performance, reliability, and maintainability.
Experience integrating AI systems with real products, databases, APIs, and operational workflows.
Strong product sense and ability to identify where AI agents create practical customer or business value.
Startup mindset: high ownership, bias to shipping, comfort with ambiguity, and ability to operate with limited direction.
Experience working in a fast-moving Seed, Series A, or similarly early-stage startup environment.

Preferred / Nice-to-have

Experience with agent frameworks such as LangGraph, LangChain, LlamaIndex, CrewAI, AutoGen, or similar.
Experience with LLM APIs such as OpenAI, Anthropic, Google Gemini, or open-source model deployments.
Experience with vector databases and search systems such as pgvector, Pinecone, Weaviate, Qdrant, Milvus, Elasticsearch, or OpenSearch.
Experience building AI evaluation pipelines, agent test harnesses, prompt regression systems, or human-in-the-loop review workflows.
Experience with backend infrastructure such as Postgres, Redis, queues/workers, event pipelines, object storage, and cloud services.
Experience with AWS, GCP, or Azure deployment patterns for Python services.
Experience building AI systems for B2B SaaS, enterprise software, infrastructure, telecom, networking, or data platforms.
Familiarity with observability tools for AI systems, including tracing, latency monitoring, cost tracking, and quality evaluation.
Experience with secure AI system design, including permissions, data access boundaries, audit logs, and safe tool execution.

Culture & Operating Principles

Ownership is real: if an agent fails, behaves unpredictably, or creates user confusion, we debug it, fix it, and harden the system.
Bias to shipping: meaningful progress weekly; fast prototypes; production mindset; tight feedback loops.
Practical AI over hype: we build systems that work reliably, not demos that only look good once.
High standards on reliability: correctness, grounding, evals, and observability matter.
End-to-end thinking: agents, backend systems, product workflows, and user experience are one system.
Low ego, high velocity: debate the idea, be direct and respectful, and keep moving.

What Success Looks Like

Multiple production AI agents are shipped and actively used across product and internal workflows.
Agent workflows are reliable, observable, and measurable through evals and execution traces.
AI features move from prototype to production quickly without becoming fragile one-off demos.
Agents can safely use tools, retrieve context, produce structured outputs, and integrate with backend services.
Evals and regression testing become part of the AI development lifecycle.
Product and engineering teams can confidently identify, build, and expand agentic AI use cases.

About Eino

Eino is building the world’s first Connectivity Digital Twin — a deeply technical, AI-driven simulation platform that models real-world environments and their connectivity layers. We solve hard, physical-world problems using advanced AI, high-performance backend systems, geometric computation, and rich large-scale data. We are a real AI company: not LLM-wrapper tooling, but deep tech. Our work spans geometry processing, spatial reasoning, GPU-accelerated simulation, data modeling, and agentic AI systems. Every engineer at Eino works on highly challenging problems and collaborates with a team of exceptional, experienced builders. We are well-funded by strong investors, have real customers, and a clear roadmap to reshape how the world understands connectivity. This is a rare opportunity for a hungry, entrepreneurial AI engineer to join a rocket-ship Seed/Series A startup at the ground level, help architect our agentic AI systems, and grow into the owner of critical AI product infrastructure.