Rounds 1–2 are fully visible · Round 3+ unlocked with a free account
MST / ClusteringGreedyRule Engine SimulationPriority Ordering
Two medium-hard problems — one algorithmic, one system simulation. The SDE-2 OA is harder than SDE-1 primarily through the follow-up constraint: optimize for scale, not just correctness. A solution that works on 100 stores needs to survive on 1,900 stores with 50M weekly transactions.
What to expect
- ›Problem 1 (Medium-Hard): e.g., 'Find the optimal set of N distribution centers to minimize average delivery distance to all stores in a region' — graph problem with greedy clustering or K-medoids reasoning
- ›Problem 2 (Medium): e.g., 'Simulate a promotional pricing engine — rules arrive in priority order, some overlap, some are exclusive; compute the final price for each cart item' — rule engine simulation with careful precedence handling
- ›Both problems must pass all test cases — partial scoring is limited on Target's OA platform
- ›90 minutes sounds generous; it isn't if you hit a tricky test case on Problem 1 and debug for 20 minutes
Practice on KnowEase.AI
Java Spring BootDesign PatternsAsync ProcessingRefactoring
Target's pair programming round at SDE-2 is deliberately close to day-to-day work. In 2024-2025 reports from the L5 Bengaluru track, candidates received a Java Spring Boot service — e.g., one that reads a large product CSV and computes store-level inventory metrics — and were asked to fix bugs, apply a design pattern (usually Strategy or Factory), implement async processing, and write tests. This is not a LeetCode session. Candidates who only practiced algorithmic problems on a blank editor are consistently caught off-guard.
What to expect
- ›A real Java codebase snippet: e.g., a Spring Boot batch job reading 500K product records from a CSV; the bug is a missing null check on discontinued items causing a NullPointerException that only surfaces in production-scale files
- ›Feature addition: 'Add asynchronous processing so this batch job doesn't block the main thread' — CompletableFuture or @Async annotation with a thread pool config
- ›Design pattern question: 'Refactor the discount calculation logic so adding a new promotion type doesn't require modifying existing classes' — Strategy pattern, with interviewers checking whether you can name it and draw the class boundary
- ›Follow-up after coding: 'How would you run this job on 50M product updates per day in near-real-time?' — they want Kafka + consumer groups, not just a bigger server
Practice on KnowEase.AI
CTEsLEAD/LAGNTILEConditional AggregationPartitioning Strategy
The SDE-2 SQL round is longer and harder than SDE-1. Three to four queries of increasing complexity on a retail schema — orders, inventory, Target Circle loyalty — followed by a schema design question and an optimization discussion. The Bengaluru team owns Target's global analytics infrastructure, so SQL is treated as a core engineering competency, not a data-analyst test.
What to expect
- ›Query 1: 'Find the top-selling product in each department for the last 30 days — but exclude any SKUs that were on promotion during that window' — CTE to build the promo exclusion list, then a NOT IN or LEFT JOIN / IS NULL exclusion pattern; candidates who use a subquery in WHERE for this get asked about performance on 500M rows
- ›Query 2: 'Calculate the 7-day rolling average of daily store revenue, and flag stores where this week's rolling average is more than 15% below last week's' — LAG() window function with conditional CASE logic; the tricky part is the date spine — stores with no sales on some days break naive window frames
- ›Query 3: 'For each Target Circle member, find the average time between consecutive purchases and rank them Frequent / Occasional / Lapsed' — LEAD() with DATEDIFF inside a CTE, then NTILE(3) or CASE on the result; candidates who get confused about which LEAD offset to use get eliminated
- ›Schema design: 'Design the tables for Target Circle loyalty points — point accrual per transaction, point expiry at 12 months, tier upgrades at 500/1000/2000 points, store-level and online accrual tracked separately'
- ›Optimization: 'Your daily revenue aggregation query on the orders table (500M rows, 3 years of data) takes 45 seconds — walk me through your indexing, partitioning, and materialization strategy'
OODDesign PatternsState MachineExtensibility
At SDE-2, Target adds an LLD round that tests object-oriented design with actual class hierarchies, not just component diagrams. Problems are retail-adjacent or classic OOD exercises. The interviewer is checking whether you can translate requirements into extensible, testable code — not just identify the right data structure.
What to expect
- ›e.g., 'Design the class structure for Target's order management system — an order has line items, each item has a SKU, quantity, and price; orders can be BOPIS (buy online, pick up in store), ship-to-home, or same-day delivery; each fulfillment type has different state transitions' — abstract Order class with concrete fulfillment subclasses, State pattern for transitions
- ›e.g., 'Design a notification service that sends restock alerts to store managers when inventory drops below threshold — different alert channels (email, SMS, push), different urgency tiers, retry logic' — Observer pattern or Strategy for channels, a Queue abstraction for retry
- ›Classic LLD as fallback: parking lot, elevator system, or library management — Target interviewers have used these, but they push for real-world complications (a parking lot with EV charging spots that have different pricing rules)
- ›Interviewers check extensibility: 'Now add a new fulfillment type — curbside pickup — without changing your existing Order class' — if your design requires touching 4 existing classes, you'll be asked to redesign
Event-Driven ArchitectureEvent SourcingReal-time Inventory SyncRecommendation SystemsA/B Testing Infrastructure
Target's system design round is deliberately practical — the interviewer said explicitly in multiple reports that they don't want rate-limiters or generic chat apps. They want retail-scale distributed problems: inventory consistency across 1,900 stores and target.com during a flash sale, personalization pipelines serving 30M weekly guests, supply chain data flows. Expect Black Friday as a scaling constraint: 5M concurrent users when the 8 AM sale opens.
What to expect
- ›e.g., 'Design Target's real-time inventory sync system — how do you keep the online product page and the in-store POS in agreement on available quantity during a flash promotion where 10,000 people are buying the same item?' — event-driven architecture with Kafka, optimistic locking vs. reservation model, eventual consistency trade-offs for UX
- ›e.g., 'Design Target Circle's loyalty points platform — point accrual per transaction (online and in-store), point expiry after 12 months, tier recalculation, fraud prevention on point manipulation' — event sourcing for the ledger, read-through cache for tier status, batch expiry job with idempotency
- ›e.g., 'Design the personalized product recommendation service for target.com' — the interviewer focuses on the real-time inference latency vs. batch model freshness trade-off, not the ML algorithm itself; they want feature store, candidate generation, ranking, A/B testing infrastructure
- ›Scale probe always comes: 'Your design works at normal load — now it's Black Friday 8 AM and you have 5x traffic in the first 10 minutes. What breaks first, and how did you design for it?'
- ›ML integration question if time permits: 'Where does your recommendation model refresh — batch nightly or near-real-time? What's the latency-freshness trade-off for a guest who just added an item to their cart?'
Business Impact FramingCross-functional CommunicationRetail Engineering Motivation
Target's Bengaluru engineering team works across merchandising, supply chain, search, and personalization — all of which require engineers who communicate technical trade-offs to non-technical product and business partners. The hiring manager round at SDE-2 tests this directly. They want engineers who understand the retail business, not just the tech stack.
What to expect
- ›Tell me about the project you're most proud of — frame the business impact first, the technical solution second; answers that lead with 'I used microservices' without quantifying customer or revenue impact get probed hard
- ›Describe a time you had to push back on a product or business request because it would have created technical debt or a reliability risk — they want to see you held the position and documented the trade-off, not just acquiesced
- ›How do you handle conflicting priorities between an engineering timeline and a merchandising team's promotional launch deadline? — they are checking whether you can negotiate scope, not just absorb pressure
- ›What excites you about building technology for a company that operates both physical stores and a digital platform at scale? — candidates who cannot give a specific answer about why the omnichannel problem is interesting rarely pass
