AI Architect

Own and define the reference architecture for FourKites AI agents, including agent loops, tool execution, state management, model routing, and evaluation frameworks.

Build and maintain both offline and online evaluation systems to measure agent performance against key business outcomes such as tracking accuracy and appointment fill rate.

Lead architecture decisions, design reviews, and mentor engineering teams while integrating agents with event-driven platforms and customer systems.

Minimum 10 years of experience building software or ML systems, with at least 3 years specifically on large language model (LLM) based AI agent systems operating in production with real user traffic.

Proficient in Python with strong hands-on coding skills; comfortable reviewing Java or TypeScript services.

Deep expertise in ML models (encoders, transformers), context engineering (token budgets, retrieval, memory), and harness engineering (agent loops, tool dispatch, state management, recovery).

Experience with cloud platforms (AWS or GCP), containers, Kubernetes, and event-driven architectures. Role based in Chennai (hybrid).

Has a proven track record operating LLM agent systems with multi-step workflows, tool integration, failure handling, and human-in-the-loop escalations under real production conditions.

Experienced in designing scalable, robust AI architectures balancing cost, latency, and quality trade-offs, and integrating complex AI agents within distributed microservices ecosystems.

Skilled at building and operating comprehensive evaluation frameworks using production data, automated graders, and feedback loops to drive continuous agent improvement and align AI performance with business metrics.