Ola & Krutrim SDE-1 & ML Engineer — Hiring Structure, Syllabus & 8-Week Prep Plan
Ola and Krutrim offer opportunities at the intersection of mobility and AI. Ola's engineering team focuses on large-scale geospatial systems, ride-matching, and pricing engines. Krutrim, India's AI unicorn, focuses on foundational LLMs, AI cloud infrastructure, and Indic-language processing. This guide provides a detailed preparation path for both software and machine learning tracks.
SDE-1 focus: DSA, geospatial systems, and distributed services. ML Engineer focus: Python, SQL, ML fundamentals, LLM fine-tuning, and mobility-specific AI cases like ETA and demand forecasting.
Last updated:
Exam Pattern
SDE-1 focus: DSA, geospatial systems, and distributed services. ML Engineer focus: Python, SQL, ML fundamentals, LLM fine-tuning, and mobility-specific AI cases like ETA and demand forecasting.
| Section | Questions | Time | Type | Difficulty |
|---|---|---|---|---|
| SDE-1 · Online Assessment | 2–4 problems | 90–120 min | DSA + Mobility Logic | Medium–Hard |
| SDE-1 · DSA Interview | 1–2 problems | 45–60 min | Live Coding & Optimization | Medium–Hard |
| SDE-1 · System Design | 1–2 cases | 45–60 min | Geospatial & Distributed Services | Hard |
| ML · Coding/SQL Assessment | Mixed | 90 min | Python, DSA, SQL | Medium |
| ML · Fundamentals Round | Varies | 45–60 min | Stats, Modeling, Evaluation | Medium–Hard |
| ML · Applied/Systems Round | 1–2 cases | 60 min | Model Training & Serving at Scale | Hard |
Section-Wise Syllabus
Everything that appears in the assessment, grouped by section.
1. SDE-1 — Core DSA & Backend
Advanced Arrays & Strings
- • Prefix sums
- • Difference arrays
- • Two pointers
- • Sliding window
- • Hashing
- • Sorting-based optimization
- • Subarrays & subsequences
- • Coordinate compression
- • Minimum window
- • String matching (KMP/Rabin-Karp)
Graphs & Mobility
- • BFS & DFS
- • Connected components
- • Cycle detection
- • Topological sort
- • Dijkstra (Route optimization)
- • Minimum spanning tree
- • DSU
- • Multi-source BFS
- • A* search basics
- • Geohashing basics
Dynamic Programming
- • House robber
- • Coin change
- • 0/1 Knapsack
- • Subset sum
- • Grid DP
- • LIS
- • LCS
- • Edit distance
- • Stock buy-sell
- • Interval DP
- • Tree DP
Backend Systems
- • REST & gRPC
- • Microservices
- • Load balancing
- • Caching strategies
- • DB Indexing & Sharding
- • Message queues (Kafka/RabbitMQ)
- • CAP Theorem
- • Idempotency
- • Rate limiting
- • Observability (Logs, Metrics, Traces)
2. ML Engineer — Modeling & Systems
Mathematics
- • Bayes’ theorem
- • Random variables
- • Expectation & Variance
- • Common distributions
- • Sampling techniques
- • Hypothesis testing
- • p-values
- • A/B testing
- • Central limit theorem
- • Correlation vs Causation
Machine Learning Core
- • Linear & Logistic Regression
- • Decision Trees & Random Forests
- • Gradient Boosting (XGBoost/LightGBM)
- • Clustering (K-Means/DBSCAN)
- • PCA
- • Feature Engineering
- • Regularization (L1/L2)
- • Cross-validation
- • Model Calibration
Deep Learning & LLMs
- • Backpropagation
- • Activation functions
- • Optimizers (Adam/SGD)
- • CNNs & RNNs
- • Transformers
- • Self-Attention Mechanism
- • Tokenization (BPE/SentencePiece)
- • LLM Fine-tuning (PEFT/LoRA)
- • RAG Systems
- • Quantization & Distillation
ML Systems
- • Data/Feature pipelines
- • Model Serving (Latency vs Throughput)
- • Online vs Batch Inference
- • Monitoring & Drift detection
- • GPU utilization
- • Vector databases
- • A/B testing for models
3. Mobility & AI Specialty
Ola Specialty
- • Haversine distance
- • Nearest-neighbor search
- • Quadtrees & R-trees
- • Spatial indexing
- • Bipartite matching
- • Demand-supply balancing
- • Surge-pricing logic
- • ETA prediction
Krutrim Specialty
- • Indic-language challenges
- • Multilingual tokenization
- • Causal language modeling
- • Safety & Red teaming
- • Large-scale inference
- • Indic-language evaluation
- • Low-resource NLP
Most Asked Topics
Highest-frequency areas across recent test papers.
Geospatial Indexing
Implementing search for nearest drivers using Geohashes or Quadtrees.
Ride-Booking State Machine
Handling transitions between requested, assigned, started, and completed states reliably.
ETA Prediction Models
Using historical traffic, distance, and real-time conditions in a regression or deep learning framework.
Transformer Complexity
Understanding O(n²d) attention complexity and its impact on long-context inference.
Idempotent Payment Processing
Ensuring ride payments are processed exactly once despite retries or network failures.
Demand Forecasting
Time-series forecasting to position drivers in high-demand areas before peak hours.
Free Resources
Curated practice links per section.
50+ DSA Practice Problems
Solve in order — arrays and strings first, then sorting and recursion.
Arrays & Strings
- Longest substring without repetition
- Maximum subarray sum
- Smallest missing positive integer
- Product of array except self
- Minimum swaps to sort
- Merge intervals
- Subarray sum equals K
- Longest consecutive sequence
Lists, Stacks & Queues
- Reverse linked list
- Reverse in groups of k
- Detect cycle and entry point
- Clone list with random pointers
- Min stack
- Next greater element
- Largest rectangle in histogram
- Sliding window maximum
Trees & Graphs
- Lowest common ancestor
- Validate BST
- Serialize & deserialize binary tree
- Number of islands
- Shortest path in weighted graph (Dijkstra)
- Word ladder
- Course schedule (Topological sort)
- Bipartite graph check
DP & Matrix
- Edit distance
- Coin change
- Longest increasing subsequence
- House robber
- Stock buy-sell with cooldown
- Search in a 2D matrix
- Set matrix zeroes
- Rotate image
Previous Year Questions & Patterns
Repeat coding questions with the expected approach.
SDE-1 (Mobility Logic)
- Design a system to find the top 5 nearest drivers to a customer's location.
- How would you handle race conditions when two customers try to book the last available cab?
- Implement a function to compute the Haversine distance between two sets of coordinates.
- Design a trip-state management service that handles network drops and app kills.
- How would you optimize route calculation for 100 delivery points (Traveling Salesman variation)?
ML Engineer (Modeling & Systems)
- How would you handle delayed labels in an ETA prediction model?
- Explain the tradeoff between precision and recall in a fraud detection system for ride cancellations.
- How would you detect data drift in a production demand-forecasting model?
- Describe how you would build a RAG system for Krutrim's customer support bot.
- What are the challenges of tokenizing Indic languages like Hindi or Tamil compared to English?
- How would you reduce the latency of an LLM inference service?
20+ SQL Query Bank
Write these out by hand — joins and subqueries are asked in both the test and the interview.
Ride Analytics
High-frequency riders
SELECT user_id, COUNT(ride_id) as ride_count
FROM rides
WHERE ride_time > NOW() - INTERVAL '30 days'
GROUP BY user_id
HAVING COUNT(ride_id) > 10;Daily revenue growth
SELECT date,
revenue,
LAG(revenue) OVER (ORDER BY date) as prev_revenue
FROM daily_stats;Driver Performance
Top drivers by rating in each city
SELECT city, driver_id, rating
FROM (
SELECT city, driver_id, rating,
RANK() OVER (PARTITION BY city ORDER BY rating DESC) as rnk
FROM drivers
) t
WHERE rnk <= 5;DBMS & Core Interview Q&A
Short answers you should be able to say out loud.
What is a Geospatial Index (e.g., GIST in Postgres)?
A specialized index used to optimize queries involving spatial data like points, polygons, and distances.
Explain Eventual Consistency in a distributed ride-booking system.
It ensures that if no new updates are made, eventually all accesses will return the last updated value, often used for non-critical data like driver profile updates.
Why use NoSQL (like Cassandra) for location tracking?
To handle the high write volume of frequent GPS updates from thousands of drivers simultaneously.
OOPs Concepts + 10 Code Snippets
Theory plus the exact snippets asked in the technical round.
Strategy Pattern
Used in pricing engines to switch between 'Standard', 'Surge', and 'Discount' pricing algorithms dynamically.
Observer Pattern
Used to notify riders, drivers, and internal systems when a ride state changes (e.g., Driver Reached).
Singleton Pattern
Often used for managing connections to a global configuration or database pool.
Simple Trip State Machine
enum TripStatus { REQUESTED, ASSIGNED, ARRIVED, STARTED, COMPLETED, CANCELLED }
class Trip {
private TripStatus status = TripStatus.REQUESTED;
public synchronized void updateStatus(TripStatus next) {
if (isValidTransition(status, next)) {
this.status = next;
notifyParticipants();
}
}
private boolean isValidTransition(TripStatus from, TripStatus to) {
// Logic to prevent illegal moves like STARTED -> REQUESTED
return true;
}
}Haversine Distance (approximate)
double haversine(double lat1, double lon1, double lat2, double lon2) {
double dLat = (lat2 - lat1) * M_PI / 180.0;
double dLon = (lon2 - lon1) * M_PI / 180.0;
lat1 = lat1 * M_PI / 180.0;
lat2 = lat2 * M_PI / 180.0;
double a = pow(sin(dLat/2), 2) + pow(sin(dLon/2), 2) * cos(lat1) * cos(lat2);
double c = 2 * asin(sqrt(a));
return 6371 * c; // Earth radius in km
}Operating System Basics
Scheduling, deadlock and memory management essentials.
Concurrency for Mobility
- • Handling thousands of concurrent GPS pings
- • Optimistic vs Pessimistic locking for cab assignment
- • Thread safety in shared driver-pools
- • Distributed locks using Redis or Zookeeper
Memory & Performance
- • Garbage collection tuning for low-latency services
- • Memory-mapped files for fast geospatial lookups
- • CPU-bound vs I/O-bound microservices
4-Week Preparation Roadmap
From zero to test-ready in 28 days.
DSA Basics & Statistics
- Arrays, Strings, Hashing
- Probability & Distributions basics
Goal: Solidify the fundamental problem-solving tools.
Advanced DSA & ML Core
- Linked Lists, Stacks, Queues
- Linear/Logistic Regression & Decision Trees
Goal: Understand core structures and basic modeling.
Graphs & Deep Learning
- BFS, DFS, Dijkstra
- Neural Networks & Backpropagation
Goal: Master graph algorithms and DL foundations.
System Design & ML Systems
- Load Balancers, Caching
- Model Serving & Pipelines
Goal: Learn how to build systems that scale.
DP & NLP/LLMs
- Dynamic Programming
- Transformers & Tokenization
Goal: Solve complex optimization problems and AI architectures.
Mobility & Applied AI Cases
- Geospatial Indexing
- ETA & Demand Forecasting cases
Goal: Deep dive into Ola/Krutrim specific business logic.
SQL, LLD & AI Infrastructure
- Advanced SQL & Window Functions
- GPU scheduling & Vector DBs
Goal: Practice data manipulation and specialized infra.
Full Mocks & Review
- 2 DSA mocks, 1 Design mock
- Behavioral prep (STAR method)
Goal: Final polish and confidence building.
Exam Day Strategy
Score targets and time allocation.
Section targets
- Online Assessment: Solve 2/3 problems fully; optimize later. Use helper functions for coordinates.
- DSA Interview: Communicate thought process clearly. Ask about scale and edge cases (e.g., zero distance).
- Design/ML Round: Start with high-level components. Discuss trade-offs (e.g., Latency vs Accuracy).
Coding time split (50 min)
- 10 min: Clarification & Approach
- 30 min: Implementation
- 10 min: Testing & Optimization
Interview Process
What follows the online test.
SDE-1 Track
- • Recruiter Screen
- • Coding Assessment
- • 2x Technical (DSA + Design)
- • Hiring Manager (Culture/Product)
ML Engineer Track
- • Resume Review
- • Coding/Math Assessment
- • ML Fundamentals Round
- • Applied ML/Systems Case Round
- • Leadership Conversation
Mock Test Plan
Four simulations to run before the real test.
Ola SDE-1 Mock
- • 3 Coding problems (1 Easy, 2 Medium)
- • 1 System Design case (Ride Matching)
Krutrim ML Mock
- • ML MCQ set (Stats/DL)
- • 1 LLM Case study (Indic RAG)
Final Checklist
Tick all of these before test day.
Recent Interview Experiences
First-hand write-ups from students who actually sat the Ola & Krutrim process.
Be the first to share a Ola & Krutrim interview experience.
Frequently Asked Questions
Is Krutrim's process different from Ola's?
Yes, Krutrim focuses more on AI infrastructure and LLM research, while Ola focuses on mobility services and geospatial backend.
How much weight is given to projects?
Highly significant. Experience with high-scale distributed systems or state-of-the-art ML models (especially Indic-language ones) is a major plus.
Want the full structured prep for Ola & Krutrim?
The Premium Vault covers Aptitude, Core CS, DSA, SQL and HR rounds with module-wise courses and drill sheets.
Related Prep
Next steps chosen from the Ola & Krutrim pattern and other Indian Unicorn recruiters.
Other Indian Unicorn companies
Flipkart Interview Prep Guide
Flipkart exam pattern, syllabus and previous year questions · ₹18–32 LPA · intern ₹80k–1.2L/mo.
Swiggy Interview Prep Guide
Swiggy exam pattern, syllabus and previous year questions · ₹16–30 LPA.
Zomato & Blinkit Interview Prep Guide
Zomato & Blinkit exam pattern, syllabus and previous year questions · ₹14–28 LPA.
Core CS subjects for this pattern
DBMS & SQL Essentials
Normalisation, keys, transactions and the SQL queries asked in written rounds.
Operating System Essentials
Processes, scheduling, deadlock and memory management in interview-ready form.
Computer Networks Essentials
OSI vs TCP/IP, HTTP, DNS and the networking questions freshers actually get.
Recommended practice path