I had interviewed with Shipsy last year in September for an SDE role. The process consisted of three rounds and was a good mix of DSA, product design, and real-world problem solving.

Stipend: 50k/ month
CTC: 14-18 LPA

Round 1: Online Assessment

The OA was quite unique compared to standard coding rounds. Instead of asking classic LeetCode problems, it assessed whether I could take an ambiguous, real-world problem, understand detailed requirements, and build a clean, correct implementation.

Every question built upon the previous one. Majorly all questions were variations of the same drone-delivery scenario, gradually increasing the complexity.

Question-wise Analysis

Q1. Locating Drop-Off Points

I was given an input grid where different color blocks represent different buildings. My task was to identify each building and mark only its center point in a new output grid using the same color.

Q2. Enhanced Drop-Off Detection
This was an extension of the previous problem.

Instead of figuring out the output myself, I had to implement its code following the detailed implementation instructions (using NumPy, finding unique colors, computing the center using the mean, etc.)

Q3.Creating Drone Paths
This question significantly increased the implementation complexity.

Each building has to be visited in a specific order. The order is provided in the last row of the grid. You need to detect all buildings, determine their surrounding boxes, and draw straight paths connecting them in the required sequence without overwriting existing buildings.

So, basically, I had to:

• Detect buildings.
• Compute bounding boxes.
• Read delivery order.
• Connect buildings.
• Respect drawing constraints.

Q4. Delivering With the Drone
This was the most advanced coding problem.

Instead of treating buildings as colored pixels, I had to model each building as an object containing its properties. Then, identify all buildings, determine the delivery sequence, and draw paths between them, where the path width depends on the size of the building.

What it was testing

• Object-Oriented Design
• Traversal (DFS/BFS)
• Large implementation tasks

This question resembled production engineering much more than an interview puzzle.

Q5. Feature Suggestions
The final question completely changed the nature of the assessment.
There was no coding involved.

Imagine you're working on Shipsy's drone delivery system. Based on aerial images, suggest new features or improvements that could make drone deliveries safer, faster, or more efficient.

What it was testing:

• Product thinking
• Creativity
• Real-world problem solving
• Understanding of logistics and AI applications

This round checked whether the candidate could think beyond algorithms and contribute ideas that improve an actual product.

Round 2: Technical Interview

The technical interview began with a discussion of my OA solutions. The interviewer asked me to explain my approaches, justify design decisions. He suggested few more optimizations and asked me to implement those on the spot.

After that, I was asked the classic DSA problem:

Largest Rectangle in Histogram

• Brute-force approach
• Why it is inefficient
• Optimal monotonic stack solution
• Time and space complexity
• Clean implementation

The interviewer emphasized understanding the reasoning behind the solution rather than simply arriving at the final answer.

Round 3: Managerial/ COO Round

The final round was conducted by the COO, Dhruv Aggrawal, and focused on product thinking and real-world problem solving.

Problem Statement
Sometimes a delivery partner marks an order as "Failed to Deliver" even though they never actually reached the customer's location.

How would you improve this process?

My Solution

• Validate that the delivery partner has reached within approximately 500 meters of the customer's location using GPS.
• Ensure at least 2–3 call attempts have been made before allowing a failed delivery status.
• If GPS indicates that the delivery partner is moving around the customer's vicinity, encourage additional delivery attempts instead of immediately marking the order as failed.

The discussion then expanded into scalability, false positives, customer experience, and operational trade-offs.

Overall Experience

Overall, I found the interview process quite new and challenging compared to other company's hiring process. The OA was particularly unique, with questions centered around drone delivery and image/grid manipulation. The later rounds checked algorithmic knowledge with practical engineering discussions and real-world logistics scenarios.

Hopefully this helps someone walking into the same interview with a bit more confidence. Good luck! 🌟