Shahbaz Hasan Raja

• AI Infrastructure — Maglev, Heimdall, RAG Pipeline. Built Drivetrain's applied-AI surface end-to-end. Maglev - a fleet of 14 cron-scheduled agents (Claude Code + MCPs) that triage Linear bugs, nudge stale reviews, scan recurring errors, and author bug-fix PRs end-to-end with a Playwright smoke-test gate. Heimdall — a stateless MCP gateway proxying 5 internal services with TTL caching and fault isolation, collapsing N point-to-point integrations into one queryable surface. RAG Pipeline — a Gitbook-backed retrieval system watching customer Slack channels, drafting humanized answers with confidence scoring, context-aware dedup, and thread-aware follow-up handling.
• Platform & Backend — RBAC, Connector Sync, Metric Cloning. Shipped the multi-tenant platform plumbing for 500+ tenants and 4K+ users. RBAC Audit Logging — end-to-end permission tracking across 13 resource types with batch DB writes, filterable changelog APIs, CSV export, and a new Security Auditor role with member-scoped read-only access. Connector Sync Scheduling — multi-schedule CRUD APIs with server-side cron generation (daily/weekly/monthly with ordinal support), row-level locking for concurrency safety, and a per-tier toll-gate that directly enabled a premium monetization path. Cross-Version Metric Cloning — transitive dependency resolution,regex-safe formula rewriting across version-specific BigQuery references, dual-context RBAC validation, and atomic session switching for safely reusing 500K+ planning variables.

• SpaceOps. Engineered and deployed the SpaceOps tool saving INR 1 Cr/year in licensing, adopted by 300 internal users across 70+ centres. Streamlined a critical API using concurrent execution with Promises, DB indexing, and hashmap optimizations — response time dropped from 4s to 800ms.
• Locations & F&B Modules. Designed a scalable Locations module using Google Maps, PostgreSQL, Redis & OpenSearch — supporting 70+ centers and 100K+ users with write-through caching. Built a community-café ordering system with UMS SSO and Server-Sent Events for real-time order tracking, serving 3K daily users.
• Help & Support Module. Engineered a GenAI LLM-based ticket classifier (OpenAI API & prompt engineering), boosting accuracy to 85% and cutting triage time by 40%. Implemented a BullMQ-driven webhook ingestion pipeline processing 10K+ Salesforce updates/month.

• Trip Lifecycle & API Migration. Owned trip management APIs (start, end, cancel) — the core logistics workflow for driver-partner operations. Refactored and migrated 8 endpoints across microservices using the Template Design Pattern, cutting response time by 50%. Hardened P0 business APIs with middleware-level validations ensuring tenant-scoped data access.
• Bidding, Rate Card & Margin Config. Built a partner bidding system with per-lane controlled rollouts and CDE visibility, a Rate Card module for digitizing commercial contracts, and a Margin Config system enabling senior leadership to set target gross margins at client-lane-vehicle granularity — directly reducing revenue leakage.
• Infrastructure & Data Pipelines. Designed a WhatsApp chatbot via Meta Webhooks for order lifecycle management. Migrated Redis to Sentinel for HA, secrets to GCP Secret Manager, decoupled FE/BE pipelines (30→5 min builds), and built Airflow DAGs warehousing 20+ MongoDB collections into PostgreSQL for BI.

Bachelor of Technology, Electronics & Computer Science — GPA 9.15 / 10