Startup · Bengaluru
CRED interviews are highly competitive, emphasising elegant code, distributed systems, and autonomous thinking.
Process: DSA Screen → Technical × 2 → System Design → HR
What is CRED's product and why does it attract high-calibre engineering talent?
Tip: CRED: credit card bill payment + rewards platform for India's creditworthy population. Engineering appeal: hypergrowth startup, hard ML problems (credit scoring, fraud), complex UX craft requirements, and a famously high engineering bar. CRED interviews are among the hardest in India — expect deep dives on every answer.
Design CRED's credit card bill payment reminder and nudge system.
Tip: Trigger: N days before due date (personalised — 3/5/7 days based on user responsiveness history). Channel selection: push → SMS → email based on click-through rates. ML model: predict optimal send time per user (open rate maximisation). Deduplication: one nudge per day per user. A/B test message variants.
How would you build a fraud detection system for CRED's high-value credit card transactions?
Tip: Features: transaction amount vs typical spend (z-score), new merchant category, unusual hour, device fingerprint change, location velocity (two cities in 10 min). Ensemble: rule engine (<1ms blocking obvious fraud) + gradient boosting model (5ms) + post-transaction async deep analysis. Feedback loop: dispute outcomes retrain the model.
What is CRED's engineering philosophy around code elegance and simplicity?
Tip: CRED is known for hiring engineers who care about craft — clean APIs, simple abstractions, readable code. In interviews: interviewers will ask you to simplify a solution you've given. Practice refactoring: after solving a problem, ask yourself 'can I make this 30% shorter without sacrificing clarity?'
Write code to implement a Least Frequently Used (LFU) cache.
Tip: Data structures: HashMap<key, value>, HashMap<key, frequency>, HashMap<frequency, LinkedHashSet<keys>> (ordered set for LRU within same frequency), min_freq counter. Get: increment frequency, update freq maps. Put: on overflow, evict min_freq key. All operations O(1). One of the hardest LeetCode Hard cache problems.
How does CRED use ML for personalised rewards and offer recommendations?
Tip: Signals: transaction history (categories, merchants, amounts), reward redemption patterns, app engagement. Models: collaborative filtering for 'users like you redeem X', content-based for new users, contextual bandits for real-time offer A/B testing. CRED's engineering blog has detailed posts on their recommendation stack.
Explain how you would optimise a slow database query in a production system.
Tip: Process: EXPLAIN/EXPLAIN ANALYZE to see query plan → identify full table scan or index miss → add index on filter/join columns → verify with staging benchmark. Advanced: covering index (includes all queried columns), partial index (where active = true), query rewrite to avoid functions on indexed columns.
Tell me about a technically challenging problem you solved that you are most proud of.
Tip: CRED interviewers probe very deeply — expect follow-ups like "why that data structure?", "what would fail at 10× scale?", "how would you change this now?". Prepare to go 3–4 levels deep on your answer. Choose a problem where you made non-obvious technical decisions you can defend.
What is functional programming and does CRED use any functional languages?
Tip: Functional programming: immutable data, pure functions (no side effects), function composition. Languages: Haskell, Clojure, Scala, Elixir. CRED uses Clojure for some backend services — a bold choice that attracts engineers who enjoy programming as a craft. Even in Java/Go, FP concepts (streams, immutable records) improve code clarity.
How would you approach building a credit score display feature that is accurate and explainable?
Tip: Credit score data comes from bureaus (CIBIL, Experian). Key engineering: parse bureau XML/JSON responses, map factors (payment history, utilisation, inquiries) to readable explanations, cache scores with 30-day TTL (bureau pulls are expensive), handle PII with field-level encryption.
CRED (Dreamplug Technologies) interviews follow a 4-round process. Here is what to expect and how to prepare for each stage.
Upload your resume and get questions scored across technical depth, communication, structure, confidence, and relevance — the same criteria CRED panels use.