Agentic Frameworks & State Management: LangGraph, CrewAI, LlamaIndex, or custom state machines; Mem0-style persistent memory layers.
LLMs & Inference Engine: OpenAI API, Anthropic Claude, vLLM / Ollama for local SLM/LLM inference execution.
Testing & Infrastructure: PyTest, benchmark test harnesses, Docker, CI/CD pipelines for regression testing.
Key Responsibilities
Stateful Traversal: Design and implement graph traversal that carries state and context forward across steps driven by user intent, avoiding reasoning restarts on each turn.
Context Modeling: Model the mathematics of context building during traversal—defining what to retain, what to prune, and how to weight paths given current intent and evidence.
State & Memory Optimization: Bridge the memory-and-logic gap so multi-turn diagnostic reasoning remains coherent, stateful, and fast enough for near real-time interaction.
Latency Reduction: Optimize traversal speed to achieve low-latency performance suitable for live, interactive environments.
Quality Assurance: Introduce a robust regression test harness to ensure algorithmic updates or alternative approaches do not break established diagnostic behaviors.
Iterative Execution: Collaborate in tight, low-overhead loops with leadership, translating verbal directions into structured, testable code.
Core Competencies & Cultural Fit
Ownership Culture: Strong self-starter who thrives in an autonomous, outcome-oriented work environment.
Remote-Native Communication: Clear, proactive communication across enterprise communication channels (Slack, Teams, Jira).
Collaborative Spirit: A pragmatic team player who values team success, constructive code reviews, and shared ownership.
Must-Have Qualifications
Graph Algorithms: Strong, demonstrable experience with knowledge graphs and graph traversal algorithms (beyond standard graph database queries).
Context & State Math: Practical understanding of building and maintaining state during traversal based on user intent.
End-to-End Delivery: Proven ability to independently drive complex workstreams from concept to deployment.
LLM & Reasoning Systems: Experience integrating LLMs into agentic or reasoning architectures with a clear understanding of statelessness and latency constraints.
Modern Agent Memory: Hands-on familiarity with advanced agent memory/context design patterns (e.g., Mem0-style architectures) beyond naive full-history prompting.
Testing & Quality: Experience setting up regression testing frameworks to maintain stability in fast-paced development cycles.
Nice-to-Have
Exposure to root cause analysis, diagnostic frameworks, or escalation engineering domains (e.g., automated support, code analysis, debugging engines).
Interest or experience with Small Language Models (SLMs) and converting raw documentation into question-answer training pairs.
Background in reasoning-heavy systems where execution logic, timing, and state matter more than static world knowledge.
Evaluation Process
A practical design discussion focused on traversing an existing diagnostic knowledge graph using memory and intent tracking.
Review of prior projects where you took an ambiguous problem to a production-grade implementation independently.
Evaluation of your approach to balancing latency, state trade-offs, and regression testing.
