Agentic Superintelligence Labs

Careers

Actively Hiring — Immediate Consideration

Agentic AI Engineer

We're hiring an autonomous systems engineer to architect, build, scale, secure, and operate production-grade agentic systems — from first principles, no guidance, zero fragility tolerance.

Apply Now

Non-negotiable depth required across all of the following

Autonomous Agent Architecture

Planner / executor / supervisor / worker / critic systems
Hierarchical multi-agent orchestration
Swarm coordination and distributed cognition frameworks
Reflection, self-correction, and retry loops
Dynamic routing and state transition design
Infinite-loop prevention and termination conditions
LangGraph — stateful, durable, long-running workflows
Human escalation and override architecture

Distributed Systems Engineering

Fault tolerance and high-availability architecture
Distributed coordination and consensus
Graceful degradation and load shedding
Backpressure handling and tail-latency optimization
Deterministic replayability and rollback systems
Idempotency and durable execution
SLO/SLA engineering and MTTR optimization
Queue-based and event-driven architectures

AI Infrastructure and ML Systems

Distributed inference systems and multi-model orchestration
Cost-aware and latency-aware model routing
Batch vs streaming inference tradeoffs
Multi-GPU serving and inference optimization
Continuous evaluation and online/offline eval systems
ML observability and training/inference tradeoffs
vLLM, TensorRT-LLM, Ray, Kubeflow, MLflow

Reasoning and Cognitive Architecture

ReAct, Chain-of-Thought, Tree-of-Thought
Deliberative inference and long-horizon planning
Self-reflection, critic models, debate architectures
Goal decomposition and constraint satisfaction
PDDL, symbolic planners, constraint solvers
Policy enforcement and autonomy boundary design

Memory, State, and Persistence

Episodic, semantic, working, shared, and task memory
Checkpointing, durable execution, and replayability
Memory compression, context pruning, retrieval ranking
Postgres, Redis, pgvector, Pinecone, Weaviate, Chroma, Neo4j
Idempotency and deterministic state recovery

Tool-Use and Enterprise Orchestration

Tool schema design and function-calling optimization
Parallel tool execution and dynamic tool discovery
Tool reliability scoring, permissioning, error-aware retries
MCP integration and agent interoperability protocols
SAP, Oracle, Salesforce, ServiceNow, Workday
Event-driven tool orchestration pipelines

Multi-Agent Coordination

Agent-to-agent protocols, message passing, consensus
Negotiation, conflict resolution, task arbitration
Blackboard architectures and distributed cognition
A2A event bus design, Redis Streams, NATS
Coordination failure modes and deadlock prevention

Context Engineering

Prompt layering and dynamic context injection
Attention budgeting and context prioritization
Semantic compression and retrieval ranking
Structured output and JSON mode mastery
Cost-aware multi-model routing: planner vs executor vs fast model

Event-Driven and Async Architecture

Kafka, RabbitMQ, AWS SQS/SNS, EventBridge
Streaming events, dead-letter queues, pub/sub
Celery, Temporal, Prefect — durable workflow orchestration
Resumable background execution under failure
Event → agent → tool → database → notification pipeline

Layer	Technologies
Orchestration	LangGraph, CrewAI, AutoGen, OpenClaw, OpenCore, Temporal, Prefect, Airflow, Dagster
LLM Layer	LangChain, LlamaIndex, Haystack, OpenAI API, Anthropic API, Open-weight inference
ML Infrastructure	Ray, Kubeflow, MLflow, vLLM, TensorRT-LLM
Tool Standards	MCP, OpenAPI, gRPC, A2A protocols, Custom event bus
Retrieval / RAG	pgvector, Pinecone, Weaviate, Chroma, Neo4j, Graph RAG, Hybrid search
Memory / State	Postgres, Redis, S3/Blob, SQLAlchemy, Alembic, Checkpointing
Distributed Systems	Kafka, RabbitMQ, SQS/SNS, EventBridge, Redis Streams, NATS, WebSockets
Durable Workflows	Temporal, Celery, Prefect, Airflow, Dagster
Evaluation	LangSmith, Ragas, DeepEval, Custom harnesses, Simulation environments
Guardrails	Guardrails AI, NeMo Guardrails, Policy checks, Sandboxing
Runtime / API	FastAPI, Pydantic, Uvicorn, asyncio, httpx, gRPC, GraphQL, Webhook
Deployment	Docker, Kubernetes, Helm, Terraform, GitOps, GitHub Actions
Cloud	AWS, Azure, GCP, IAM / RBAC, Serverless, Cloud-native infrastructure
Observability	OpenTelemetry, Datadog, Prometheus, Grafana, ELK stack, LangSmith
Security	OAuth / JWT, Vault, AWS Secrets, Azure Key Vault, RBAC, Sandboxing
Enterprise APIs	Salesforce, ServiceNow, SAP, Oracle, Workday, Industrial systems
Languages	Python (async mastery)
Formal Planning	PDDL, Symbolic planners, Constraint solvers

Layer

Technologies

Orchestration

LangGraph, CrewAI, AutoGen, OpenClaw, OpenCore, Temporal, Prefect, Airflow, Dagster

LLM Layer

LangChain, LlamaIndex, Haystack, OpenAI API, Anthropic API, Open-weight inference

ML Infrastructure

Ray, Kubeflow, MLflow, vLLM, TensorRT-LLM

Tool Standards

MCP, OpenAPI, gRPC, A2A protocols, Custom event bus

Retrieval / RAG

pgvector, Pinecone, Weaviate, Chroma, Neo4j, Graph RAG, Hybrid search

Memory / State

Postgres, Redis, S3/Blob, SQLAlchemy, Alembic, Checkpointing

Distributed Systems

Kafka, RabbitMQ, SQS/SNS, EventBridge, Redis Streams, NATS, WebSockets

Durable Workflows

Temporal, Celery, Prefect, Airflow, Dagster

Evaluation

LangSmith, Ragas, DeepEval, Custom harnesses, Simulation environments

Guardrails

Guardrails AI, NeMo Guardrails, Policy checks, Sandboxing

Runtime / API

FastAPI, Pydantic, Uvicorn, asyncio, httpx, gRPC, GraphQL, Webhook

Deployment

Docker, Kubernetes, Helm, Terraform, GitOps, GitHub Actions

Cloud

AWS, Azure, GCP, IAM / RBAC, Serverless, Cloud-native infrastructure

Observability

OpenTelemetry, Datadog, Prometheus, Grafana, ELK stack, LangSmith

Security

OAuth / JWT, Vault, AWS Secrets, Azure Key Vault, RBAC, Sandboxing

Enterprise APIs

Salesforce, ServiceNow, SAP, Oracle, Workday, Industrial systems

Languages

Python (async mastery)

Formal Planning

PDDL, Symbolic planners, Constraint solvers

Every single one of the following must be true

Production track record

You have built and operated autonomous or multi-agent systems in real production environments with real uptime, reliability, and operational consequences — not prototypes, not demos.

Full-stack depth, zero gaps

You independently own architecture, orchestration, infrastructure, deployment, observability, security, evaluation, and operational reliability end-to-end. No "I'd delegate that."

Distributed systems first

You think in distributed systems, reliability engineering, and execution guarantees before writing a line of code.

Autonomous without scaffolding

You can design full autonomous architectures independently from first principles without guidance, supervision, or a technical lead.

Security-first by default

You build security-first systems from day one — attack surface, prompt injection, privilege escalation, and audit requirements are architectural inputs, never afterthoughts.

Evaluation before scaling

You build evaluation systems before scaling agents. You measure task success, planning quality, tool correctness, hallucination rate, and cost efficiency.

Long-running workflow mastery

You have built and operated autonomous workflows that execute over hours or days — with durable state, mid-execution failure recovery, and deterministic replayability.

Deep async Python mastery

asyncio, context managers, generators, concurrency, queues, dependency injection — fluently and deeply.

Quantitative reliability reasoning

You reason quantitatively about latency, reliability, throughput, and cost tradeoffs. You have SLO targets and you hit them.

Enterprise integration fluency

You have integrated at least two of Salesforce, SAP, ServiceNow, Oracle, or Workday in an agentic or automated production pipeline.

Human-agent collaboration design

You architect escalation paths, confidence thresholds, interruptibility, and explainability from first principles.