Agentic AI Engineer (Cheminformatics) в Software Country (ТОО Балхаш Системс) Сербия

We have 30 years of expertise in designing and building custom software systems. We provide software development services focusing on complex high-load applications, AI and BI solutions, and mobile apps.

Project focused on building an agentic assistant for setting up, running, and interpreting complex workflows in early-stage drug discovery.

The goal is to improve how scientists initiate and manage these workflows by replacing long manual setup forms with a conversational assistant. The assistant works alongside scientists to gather the needed inputs from existing data sources and documents, tracking the workflows as they progress, and answering questions along the way while keeping scientists in control of key decisions.

The scope of work includes building and orchestrating the agents, skills, and workflows that drive the process, running predefined and custom analyses on the fly, and developing the underlying data layer (memory, state, retrieval, and decision records). The assistant must support multiple concurrent workflows ​and handle context across users and sessions. In production we must persist and retrieve state reliably, enforce cost and safety guardrails, and keep decisions traceable and auditable.

Functionality will be integrated with existing molecular design and data-driven decision-making tools.

Core Responsibilities

• Design and implement agentic AI systems, including agent workflows and orchestration
• Design and own the data structures behind agentic pipelines (memory, state, retrieval, tool use) in production
• Build and iterate on prompts, skills, and agent definitions
• Integrate agents with APIs and databases
• Collaborate with the front-end engineer to integrate agent interactions into a chat interface

Required Experience

• Strong understanding of a variety of LLMs and their capabilities and agentic infrastructure
• Experience designing and integrating APIs for AI-powered systems
• Comfortable working across the stack — from prompt engineering to data layer to UI handoff

Nice to Have

• Hands-on experience with Amazon Bedrock
• Familiarity with multi-agent frameworks and patterns
• Experience with evaluation and observability for agentic systems
• Prior work integrating agents into conversational UIs
• Experience in chemistry, cheminformatics, or drug discovery domain
• Python

Our offer as your future employer

• Collaboration via a B2B contract with payments in USD or EUR, depending on your preference, or through a labor contract
• Flexible work schedule.
• Possibility to work remotely
• Opportunities for professional growth.
• A company laptop to ensure a comfortable and efficient work setup.​​​​​