Sreya Vallabhaneni

About

Built and shipped an AI-driven interview evaluation pipeline at SeekOut that automated recruiter screening via a multi-stage LLM agent workflow (.NET backend, RabbitMQ orchestration, Python workers). Emphasizes production-grade reliability (idempotency, retries, strict JSON/schema validation), strong observability with OpenTelemetry, and measurable efficiency gains including ~40% reduction in token usage/cost.

Key Strengths

• Shipped production LLM agent pipeline for interview evaluation at SeekOut
• Designed fault-tolerant async architecture (.NET + RabbitMQ + Python workers)
• Improved reliability via schema validation, retries, and idempotency
• End-to-end observability using OpenTelemetry tracing and stage-level metrics
• Reduced LLM token usage/cost by ~40% through prompt optimization and structured outputs
• Built scalable multi-stage orchestration with persisted intermediate state and stateless workers
• Implemented offline eval datasets and production monitoring/feedback loops to iterate on reliability
• Handled messy multi-source candidate/job data with normalization, safe defaults, and guardrails
• End-to-end ownership of full-stack features (frontend + backend + post-launch iteration)
• Debugging and resolving stale/partially-updated UI state across multiple backend services
• Frontend performance tuning by reducing unnecessary re-renders via component decomposition and localized state
• Postgres query optimization using composite indexes and EXPLAIN plan validation
• Designing resilient async/event-driven workflows with retries, idempotency, and reconciliation
• Shipping quickly in high-growth startup environments while adding production safeguards (validation, logging)

Experience

Education

Skills

Languages