Bank Of America Merrill Lynch Software Engineer Interview Guide

Three rounds. Java was easy—my prep paid off. Then came the whiteboard: mapping a Node.js schema. I flatlined, then pivoted hard. I left fried. You don't conquer; you survive a corporate street fight.

Questions asked:

Round 1: Java & Spring Boot Core

• Concurrency & Memory:

• "How does ConcurrentHashMap achieve thread safety in Java 8+ compared to Java 7's segment-locking approach?" (Looking for CAS operations, synchronized on bucket head nodes, and red-black tree conversion under high collision).

• "Walk me through how Garbage Collector (specifically G1 or ZGC) identifies GC Roots and manages the evacuation phase without causing prolonged Stop-the-World pauses."

• "What actually happens when a Virtual Thread (Project Loom) blocks on I/O vs when a platform thread blocks? When does thread pinning occur?"

• Spring Boot Framework Internals:

• "If you call a @Transactional method internally from within the same class using this.myMethod(), why does the transaction fail to start? How do Spring’s CGLIB vs JDK dynamic proxies handle this?"

• "By default, @Transactional only rolls back on unchecked exceptions (RuntimeException). How do you handle rollback for checked exceptions or custom domain errors without polluting business logic?"

Round 2: Architecture, Schema Design & Resiliency

• The Main Schema & Workflow Prompt:

• "Scenario: A user changes their phone number on the profile service. However, three downstream services (Auth/2FA, Order Notifications, and Marketing) store indexed phone records. Design a phone number disassociation workflow."

• Follow-up: "Draw the PostgreSQL schema for the user_phone_mappings table. How do you handle unique constraint indexes when soft-deleting a disassociated phone number so the same number can be immediately re-bound by another user?"

• Edge/Node.js component: "If the ingress gateway uses a Node.js edge worker for lightweight token validation, how do you handle asynchronous event propagation down to Java/Spring Boot microservices without losing context?"

• Cloud Control Plane & Resiliency (AWS focus):

• "If the downstream Notification service in AWS starts throwing 503 errors during a traffic spike, how do you implement Circuit Breaking (e.g., Resilience4j) and Dead Letter Queue (DLQ) retry strategies so the primary API doesn't fail the user’s request?"

• "How do you ensure idempotency across asynchronous event consumers when processing phone disassociation events out of order?"

Round 3: Lead-Level Scenarios & Production Diagnostics

• System Failure & Debugging:

• "Your Spring Boot service is suddenly hanging in production. CPU usage is low, but thread pool utilization is at 100%. Walk me through your first 15 minutes of diagnostics using thread dumps (jcmd/jstack) or heap analyses."

• "How do you detect and resolve connection pool starvation in HikariCP under heavy load?"

• Team Leadership & Architecture Governance:

• "You are migrating a legacy monolith REST controller to microservices. Half the team wants to do a big-bang rewrite; the client wants zero downtime. How do you pitch and execute the Strangler Fig pattern to stakeholders?"

• "How do you handle a scenario where a junior engineer continuously bypasses architectural guidelines to hit aggressive sprint delivery dates?"