C++ Algorithm Developer for Logistics in Lavka

Yandex Lavka is a service for fast delivery of groceries and everyday goods. To get an order to a user within 15 minutes after placing it, complex algorithms work behind the scenes.

The Algorithmic Services group works on discrete optimization tasks in logistics and warehouse processes. We solve NP-hard problems of routing, batching, and planning, and quickly turn ideas from scientific papers into production code.

Our team consists of engineers who:

• develop and improve the algorithmic core of logistics services in C++ (userver);
• implement heuristics and optimization algorithms for routing, batching, and planning tasks;
• transfer ideas from scientific papers to production and test them on real data;
• participate in discussions about solution architecture and the development of the algorithmic platform.

What tasks await you

Order Dispatch We plan routes so that ice cream arrives cold, coffee arrives hot, and the courier doesn't have to return to the warehouse if orders are in neighboring buildings. The algorithms combine orders into batches, select a courier, and build a delivery route in real time.

Optimization of Order Assembly in the Warehouse Orders must be ready by the time the courier arrives to pick them up. If they are assembled too early, they will wait for delivery. If too late, the courier will be idle. We build algorithms that manage the assembly queue and synchronize the work of the warehouse and delivery.

Balancing Demand and Delivery Capacity (Surge) Sometimes the number of orders exceeds what the warehouse and couriers can handle. For example, due to bad weather or a sudden spike in demand. Our algorithms regulate delivery conditions in real time to maintain service speed and not overload the system.

Choosing the Delivery Method Sometimes an order will be delivered faster by a courier on a bicycle, sometimes by car, and sometimes by a rover robot. The algorithms choose the optimal delivery type for each order, considering time, cost, and the available fleet.

More about backend at Yandex — on the channel Yandex for Backend

We expect you to

• Be proficient in C++ programming
• Have a good knowledge of algorithms and data structures
• Understand the basics of probability theory and mathematical statistics
• Enjoy delving into complex tasks

It will be a plus if you

• Have solved discrete optimization problems (VRP, scheduling, matching)
• Have worked with heuristics and metaheuristics (LNS, simulated annealing, etc.)
• Are interested in applied research