Agentic AI Engineer

Mandatory Skills & Competencies:

First Principles Thinking

• Ability to break down any agent task such as planning, memory, tool usage, and reflection into fundamental reasoning steps without relying solely on framework defaults.

System Design

• Ability to architect agentic systems from scratch, including state management, failure recovery, observability, scaling agent loops, and defining human-in-the-loop boundaries.

Roles & Responsibilities

• Design and develop multi-agent systems for business use cases from scratch.
• Architect and manage agent workflows, including planning, execution, memory, and reflection loops.
• Handle state persistence using PostgreSQL as the source of truth.
• Implement agent loops with retries, tool call timeouts, and error recovery mechanisms.
• Design memory architecture using PostgreSQL and pgvector, including session memory, episodic memory, and semantic search capabilities.
• Build and manage tool integrations, function calling, and parallel tool execution.
• Develop observability mechanisms for monitoring agent behavior and performance.
• Plan and execute evaluation strategies for agent success rates, token efficiency, and drift detection.
• Collaborate with engineering teams to deploy and scale agentic systems effectively.

Required Skills & Qualifications

• 4–6 years of relevant experience in AI, Machine Learning, or Software Engineering.
• Hands-on experience building production-grade agent loops using LangGraph or custom LLM orchestration frameworks.
• Strong expertise in PostgreSQL, including window functions, CTEs, indexes, and JSONB.
• Experience with pgvector for storing, indexing, and retrieving embeddings within PostgreSQL.
• Strong understanding of vector search architecture, including IVFFlat, HNSW, and hybrid search (full-text + vector search).
• Experience with tool usage patterns, function calling, parallel tool execution, and error recovery.
• Hands-on exposure to LLMs such as GPT-4, Claude, Gemini, Llama 3, Mistral, or similar models.
• Working knowledge of Docker and containerized deployments.
• Strong problem-solving abilities and system design capabilities.

Good to Have

• Basic exposure to Kubernetes, including deployment of stateless agent services, ConfigMaps, and liveness probes.
• Experience in JSON Prompt Engineering, including schema enforcement, constrained decoding, and self-correction techniques.
• Active GitHub contributions demonstrating continuous learning and experimentation.
• Relevant certifications related to AI, LLMs, Agentic AI, Data Engineering, or Cloud technologies.
• Experience optimizing context utilization and managing token budgets efficiently within LLM constraints.