Manohar Vallabi

About

LLM/agentic-systems practitioner who has repeatedly taken LLM-driven pricing/decision services from prototype to production using pilots, guardrails, observability, and staged rollouts. Demonstrates strong real-time incident troubleshooting (dependency timeouts, cached fallbacks) and post-incident hardening (isolation/async/alerts), and also supports go-to-market via developer workshops, technical demos, and sales-aligned POCs.

Experience

Education

Key Strengths

• Took an LLM-driven pricing/decision prototype to production via pilot validation, hardening, and staged rollout
• Defines and aligns customer success metrics (latency, business-rule accuracy, fallback rate, revenue lift/acceptance)
• Strong production reliability mindset (guardrails, monitoring, feature flags, safe fallbacks)
• Real-time debugging using observability signals and end-to-end request tracing across prompts/tools/downstream services
• Handled peak-traffic incident by identifying downstream dependency bottleneck and stabilizing with cached fallback + tighter timeouts
• Improves systems post-incident with isolation, async calls, and dependency health alerts
• Effective at tailoring technical demos/workshops to mixed seniority audiences with runnable examples and follow-up docs
• Partners with sales/customer-facing teams on discovery, demos, integration/scalability Q&A, and POCs to drive adoption
• Built end-to-end PDF Q&A app with RAG pipeline (extraction, chunking, embeddings, retrieval, contextual answers)
• Performance optimization for LLM/RAG systems (embedding caching, chunk-size/overlap tuning, parallel processing)
• Designed high-throughput Spring Boot microservice consuming millions of Kafka trading events with low latency
• Production reliability/observability for data pipelines (consumer lag, latency/throughput/error metrics, alerts, health checks, retries, circuit breakers)
• Database performance tuning in PostgreSQL (query optimization, indexing, stored procedures) for fast enrichment
• Full-stack integration of React dashboards with Java APIs; resolved real-time UI lag via payload reduction and React render optimization