Meesho Interview Questions 2026: SDE, Data & PM Roles

Meesho's SDE interview process for 2–5 year candidates runs 4–5 rounds.

Round 1: Online Assessment (HackerRank)

• 2 DSA problems, 90 minutes
• Medium to Hard difficulty: the bar is higher than Nykaa or MMT
• One problem often involves a graph or DP with logistics framing ('minimum delivery cost', 'optimal warehouse assignment')

Round 2: Technical Interview 1 (DSA deep-dive)

• 1–2 problems live on shared editor
• Interviewer probes complexity: 'Can you do better than O(n²)?' is common
• Meesho framing: catalogue ranking, seller scoring, order batching
• You'll be asked to trace through your code with a specific input

Round 3: Technical Interview 2 (System Design)

• For 3+ years: High-Level Design
• Design Meesho's product catalogue: unstructured, seller-uploaded, 100M+ SKUs
• Design the order batching and delivery partner assignment for Tier-2 India
• Design a price discovery system for resellers setting margins
• Design Meesho's notification system (WhatsApp + push + SMS at 50M users)
• For 1–3 years: Low-Level Design + data structures trade-offs

Round 4: Engineering Manager / Director

• Deep-dive on your most complex past system
• Meesho-specific culture: 'democratising commerce for Bharat': do your values align?
• Behavioural: ownership examples, cross-functional work with product and ops

Round 5: VP Engineering (for senior roles)

• Org-level thinking: how do you scale a team, not just a system?
• Architecture trade-offs at the strategic level

Meesho's DSA bar is LeetCode Medium to Hard. Social commerce framing wraps standard algorithms.

Q1: Minimum cost to connect all warehouses (Meesho logistics framing) Given N warehouses and M potential roads between them with costs, find the minimum cost to connect all warehouses. Solution: Minimum Spanning Tree: Kruskal's (sort edges + union-find) or Prim's. ```python def mincostconnectwarehouses(n, connections): connections.sort(key=lambda x: x[2]) # sort by cost parent = list(range(n + 1)) def find(x): while parent[x] != x: parent[x] = parent[parent[x]] x = parent[x] return x def union(x, y): px, py = find(x), find(y) if px == py: return False parent[px] = py return True cost = edgesused = 0 for u, v, w in connections: if union(u, v): cost += w edgesused += 1 if edgesused == n - 1: return cost return -1 ```

Q2: Rank sellers by multi-criteria scoring Given sellers with (orders, returnrate, rating, deliverytime), rank them by a composite score where weights are dynamically provided. Solution: Custom sort with weighted scoring function. Ask about tie-breaking rules: interviewers value this follow-up.

Q3: Find the minimum number of delivery trips Given a list of orders with weights and a truck capacity, find the minimum trips to deliver all orders (each trip carries a subset summing to ≤ capacity). Solution: Greedy (sort orders, two-pointer to pair smallest with largest) for approximation. DP for exact minimum with small N.

Q4: Detect duplicate product listings Given 10M product titles, find near-duplicate listings (same product uploaded by different sellers with minor title variations). Solution: Shingling + MinHash for LSH (Locality Sensitive Hashing): this is a systems design question disguised as DSA. Discuss at a high level; the interviewer tests your familiarity with approximate similarity at scale.

Q5: Optimal discount allocation Given a fixed discount budget and N sellers, allocate discounts to maximise total GMV uplift. Each seller has a (discountpct → expectedgmv_uplift) function. Solution: Greedy: sort by marginal uplift per rupee; or DP knapsack if functions are step-wise.

Meesho system design is unusual: Tier-2 India constraints apply everywhere. Low-end Android devices, 2G/3G intermittent connectivity, UPI-first payments, and WhatsApp as a primary communication channel shape the architecture.

Question 1: Design Meesho's product catalogue (100M+ unstructured SKUs)

The unique challenge: Meesho allows any seller to upload any product without a fixed taxonomy. 100M+ SKUs are seller-created, inconsistently named, and may duplicate.

Key components:

• Ingestion: Seller uploads photo + title + price. OCR extracts text from images; ML model classifies into category tree.
• Deduplication: MinHash + LSH to find near-duplicate listings. Cluster duplicates under a canonical product; show all seller offers under one PDP.
• Search: Elasticsearch with seller-provided title + ML-extracted attributes. Ranking: seller trust score + price + delivery speed.
• Cold start: New seller has no trust score → show their products lower until 10+ orders establish a baseline.

Question 2: Design the notification system for 50M users

Notification channels in priority order for Meesho's user base: WhatsApp (highest engagement for Tier-2 sellers), Push notification (Android), SMS (fallback for no internet).

• Fan-out: For a flash sale, 50M notifications must go out within 10 minutes. Use SQS (or Kafka) with 100 worker nodes, each processing 500 notifications/second.
• Template personalisation: {seller_name} started a sale. Message assembled at worker, not at producer.
• Delivery tracking: Log sent + delivered + opened per channel. Feed back to ML model to select best channel per user.
• Compliance: TRAI regulations limit SMS marketing to 10am–9pm IST. Build a scheduling layer with timezone-aware delivery windows.

Question 3: Design the order batching system for Tier-2 delivery

In Tier-2/Tier-3 cities, Meesho uses a hub-and-spoke model: orders from a pin code are batched and delivered by a single delivery partner.

• Batching window: Collect all orders for a pin code in a 4-hour window before dispatching a delivery partner.
• Assignment: Match batches to available delivery partners by proximity and capacity.
• Intermittent connectivity: Delivery partner app must work offline: sync when connectivity returns. Use optimistic updates with offline-first architecture (orders stored locally, synced on reconnect).

Meesho is solidly Tier-1 Indian product company compensation: comparable to Swiggy and Zomato, above Nykaa and MMT.

SDE: | Level | Salary Range | |---|---| | SDE-1 (0–2 yrs) | ₹20–35 LPA | | SDE-2 (2–5 yrs) | ₹35–65 LPA | | Senior SDE (5–8 yrs) | ₹60–100 LPA | | Staff / Principal | ₹90–130+ LPA |

Data Science / ML: | Level | Salary Range | |---|---| | DS-1 / MLE-1 (0–2 yrs) | ₹18–32 LPA | | Senior DS / MLE (3–6 yrs) | ₹30–65 LPA | | Lead MLE | ₹60–95 LPA |

Product Manager: | Level | Salary Range | |---|---| | APM / PM-1 | ₹22–38 LPA | | PM-2 / Senior PM | ₹38–72 LPA | | Group PM / Director | ₹72–110 LPA |

Compensation notes:

• Pre-IPO: ESOPs not yet liquid; paper value based on last valuation round (SoftBank-led Series F)
• ESOP vesting: 4-year cliff with 1-year vesting schedule standard
• Joining bonus: common for lateral hires with notice period obligations
• Location: Bengaluru only (headquarters); no Delhi or Hyderabad offices for engineering
• Negotiation: Meesho matches competing offers from Zomato, Swiggy, and PhonePe: bring a competing offer for base negotiation