AI Assistant Developer for Managers

AI Assistant Developer for Managers

Aima is an AI assistant for managers, built on an LLM foundation. Our team integrates LLM with managers' work tools: connecting calendar, task tracker (Yandex Tracker), messengers, and Wiki. We process data from various sources, extract insights, and automate routine tasks — from chat analysis to ticket creation.

We are looking for a senior backend developer. You will be responsible for the product's technical architecture, ensuring its reliability and scalability.

What tasks await you

Architecture Revision The product has moved beyond the MVP stage. Many integrations with the services used by managers are planned in the near future. The MVP architecture was focused on development speed, but now a foundation for scaling is needed.

Optimization of work with userbots in Telegram Currently, for userbots to work, at least one physical account is required per pod, which creates problems with reliability and code scalability. You will work on making the number of userbots independent of the number of pods.

LLM Tool Re-call The agent uses Tool/Functional calling in its work. Sometimes the model decides that calling a tool is unnecessary, even if the instructions state otherwise — this happens when the context is too large or it already contains an answer to the user's question. It is necessary to teach the model to re-call tools, without pulling old information from history, while preserving the dialogue context.

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

We expect you to

• Have five or more years of development experience
• Have designed reliable and scalable architecture
• Work with the product: formulate goals, test hypotheses, think about the target process

It will be a plus if you

• Know Python
• Understand the principles of LLM operation
• Have implemented LLM into production processes
• Have worked with Function Calling or Tool Calling