Pranav Chand

About

Backend engineer focused on multi-tenant enterprise AI personalization and recommendation platforms, combining ML/LLM intent extraction with deterministic policy guardrails for compliance and auditability. Has hands-on AWS experience (ECS/Lambda/DynamoDB/S3) and led a careful DynamoDB single-table migration using dual write/read, canary + feature-flag rollouts, and strong observability/security (JWT/OAuth2, RBAC, Postgres RLS).

Experience

Education

Key Strengths

• Designed and evolved a multi-tenant AI personalization/recommendation backend with strict tenant isolation
• Balances ML/LLM intelligence with deterministic, auditable policy guardrails for compliance
• Low-latency architecture optimization by splitting hot-path inference from async enrichment and caching state
• Led high-risk data migration to DynamoDB single-table design using dual write/read, shadow mode, and tenant-by-tenant feature-flag rollout
• Strong production security implementation: JWT/OAuth2, RBAC, tenant scoping, and Postgres RLS to prevent cross-tenant leakage
• Reliability engineering mindset: idempotency, conflict handling for duplicate/out-of-order events, DLQs, reconciliation to prevent silent corruption
• Operational excellence: observability with structured logs, tracing, metrics (SLIs/SLOs), and audit logs
• Built and deployed an enterprise internal GenAI platform used by multiple product teams
• Designed multi-provider LLM orchestration with vendor abstraction to avoid lock-in
• Implemented high-availability patterns (policy routing, fallbacks, retries/backoff, timeouts, circuit breakers) to handle provider degradation
• Production-grade AWS orchestration for failure-prone AI workflows (Step Functions + Lambda + EventBridge + SQS)
• Strong evaluation discipline: offline evals, integration tests, and production A/B tests/staged rollouts
• Operational rigor with measurable reliability/cost metrics (success rate, fallback rate, p95 latency, cost per successful task) and drift monitoring
• Effective cross-functional delivery: translated non-technical safety/reliability concerns into measurable acceptance criteria and operationalized launch