Networks AI Research Engineer

Role Overview

Join Banyan's core research group to build the next generation of evaluation and reasoning systems for network AI agents. You'll design experiments, craft bespoke eval suites, and push the limits of agent reliability across WiFi, cellular, and optical networks.

Responsibilities

• Design and maintain golden datasets plus eval harnesses for RCA, alarm correlation, and troubleshooting agents
• Prototype DSPy/LangGraph workflows that blend multi-hop reasoning with deterministic API/tool calls
• Create prompt optimization loops that shrink hallucinations and improve command safety
• Instrument Langfuse/MLflow pipelines to benchmark agent regressions across models and prompts
• Collaborate with network SMEs to encode domain constraints, playbooks, and safety policies into scoring logic
• Publish internal research notes on evaluation methodology, prompt tuning insights, and agent failures
• Partner with product/engineering to productionize promising research outcomes

Qualifications

• 4+ years in applied ML, LLM research, or agentic systems
• Hands-on experience with LangChain, LangGraph, DSPy or similar orchestration frameworks
• Deep understanding of evaluation methodologies, including human+AI review loops and automated scorers
• Strong Python skills plus familiarity with PyTorch/JAX or comparable frameworks
• Comfort working with network telemetry, OSS/BSS APIs, or large operational datasets
• Ability to translate ambiguous operator requirements into concrete research plans
• Excellent written communication for experiment readouts and design docs

Nice to Have

• Experience integrating agents with NetOps stacks (ServiceNow, Grafana, Kafka, Juniper/Cisco APIs)
• Background in telecom, ISP, or large-scale infrastructure troubleshooting
• Familiarity with MLflow, Langfuse, or custom eval infra
• Contributions to open-source agent frameworks or evaluation tooling
• Graduate research (MS/PhD) in CS, EE, or related field