Raunak Singh

The data-to-action layer of the platform: warehouse analytics turned into shipped features, faster reporting, and agent-ready developer tooling.

• Architected a Smart Suggestions engine mapping analytics signals to actionable promotions and catalog updates via conditional API orchestration; integrated OpenAI on Cloud Run with deterministic rule-based fallback and full audit tracking.
• Reduced analytical API latency from 30 to 60 seconds down to milliseconds via a BigQuery to Postgres hybrid architecture; built an NLP-to-SQL reporting engine with scheduled email delivery and a calendar-style UX.
• Built Zen CLI, an automation CLI unifying local-runtime orchestration for 24+ microservices and one-command Jenkins deployments (Playwright + Chrome CDP); exposed it as an MCP/Skill agent control plane so AI agents self-bootstrap service context, plans, and startup.

The ML engine behind every storefront recommendation surface, serving 1000+ sales channels.

• Designed an ML-powered recommendation system (Vertex AI, Redis, Postgres, MongoDB) delivering similar, trending, and bought-together results across 1000+ sales channels with currency and locale-aware caching.
• Replaced third-party catalog ingestion with an internal GCS-based workflow, eliminating an external dependency and reducing costs by roughly ₹3L per month.
• Scaled the engine to 9+ recommendation types (similar, trending with category, brand, and gender variants, frequently-bought-together, cross-sell, personalised, local-first recently-viewed) with request-time rank-boost and brand-diversity controls, plus an OpenAI gpt-4o-mini category-intelligence layer.

Drove the lender and merchant panels of a microservice-based fintech lending ecosystem, end to end.

• Led full-stack delivery as DRI for lender and merchant panels within a microservice-based lending ecosystem, driving configuration, transaction, reporting, role-management, and payment workflows end to end.
• Strengthened core transaction and reconciliation services (Node/Express) by refining models, cron jobs, lender-aware state handling, and summary APIs, improving correctness and operational reliability.
• Enhanced auth, session, and RBAC flows across distributed services (stonewall, enclose) by fixing token handling, route ordering, and lender-user mappings, and aligning contracts via a generated SDK layer.
• Improved platform resilience through Kafka-based async workflows, Redis caching, Postgres persistence, and Swagger-driven contracts, stabilizing cross-service communication and notification and reporting infrastructure.