Raahul Sallagunta

About

Backend and distributed-systems engineer with experience integrating LLM capabilities into clinical data workflows at CVS. Stands out for treating AI as an engineering accelerator rather than a shortcut, with strong emphasis on validation, observability, Kafka-based async pipelines, and safe multi-agent orchestration for production systems.

Key Strengths

• Strong human-in-the-loop approach to AI-assisted engineering
• Deep backend and distributed systems expertise
• Productionizing LLM workflows with validation and reliability safeguards
• Design of asynchronous Kafka-based AI pipelines
• Multi-agent workflow design with clear role separation
• Focus on architecture, security, scalability, and observability
• Operational mindset for AI systems in production
• Hands-on experimentation with emerging AI frameworks and patterns
• Technical leadership in orchestrating AI components safely
• Shipped end-to-end LLM-powered clinical note summarization feature (backend, data model, and UI)
• Designs secure APIs with OAuth2/JWT and role-based authorization for sensitive clinical data
• Builds reliable async architectures (Kafka/workers) to handle long-running processing without blocking UI
• Improves LLM output consistency via structured prompts, JSON validation, and post-processing checks
• Strong product-focused application engineering (API contracts, edge cases, clinician-facing UX)
• PostgreSQL schema design and performance tuning (indexes, composite indexes, EXPLAIN-driven optimization)
• Production AWS/EKS operations with CI/CD and observability; handled latency/scaling incident end-to-end
• Evolved synchronous LLM summarization APIs into async job-based architecture to reduce timeouts and latency spikes
• Designed reliable async processing with idempotent job IDs, retry-safe consumers, and explicit job state transitions
• Improved API responsiveness via Redis caching and reduced key API response times by ~28%
• Strong security implementation for sensitive clinical data (OAuth2/JWT + server-side RBAC, auditable access patterns)
• Proactive observability improvements (correlation IDs, queue depth/lag metrics) to detect hidden async failure modes early
• Incremental/strangler-style migration approach with backward-compatible contracts and gradual traffic rollout

Experience

Education

Skills

Languages