The AI Interview - Master AI/ML Interviews

Implement a coordinator agent for multi-agent systems:

register_specialist(name, capabilities, agent_fn): Add specialist
- capabilities: list of strings (e.g., ['math', 'code'])
route_task(task_description): Select best specialist
- Match task keywords to capabilities
- Return confidence score and rationale
coordinate(complex_task): Orchestrate multiple specialists
- Split task by 'and', 'then'
- Route each to appropriate specialist
- Aggregate results

Routing Logic: Score = count of overlapping keywords between task and capabilities

Coordination: Execute subtasks sequentially, pass context between them.

Examples

Example 1:

Input:

coord = CoordinatorAgent(); coord.register_specialist('math', ['math', 'calc'], lambda x: x); r = coord.route_task('calculate 2+2'); r['specialist']

Output: 'math'

Explanation: Task contains 'calculate', matches math capability

Starter Code

class CoordinatorAgent:
    """
    Coordinator agent that manages multiple specialist agents.
    """
    
    def __init__(self):
        self.specialists = {}  # name -> agent_config
        self.task_history = []
    
    def register_specialist(self, name, capabilities, agent_fn):
        """Register a specialist agent"""
        # Your implementation here
        pass
    
    def route_task(self, task_description):
        """
        Route task to most appropriate specialist.
        Returns {'specialist': name, 'confidence': float, 'rationale': str}
        """
        # Your implementation here
        pass
    
    def coordinate(self, complex_task):
        """
        Break complex task and coordinate specialists.
        Returns {'subtasks': [...], 'results': {...}, 'integrated_result': ...}
        """
        # Your implementation here
        pass
    
    def _match_capabilities(self, task, capabilities):
        """Score how well capabilities match task"""
        # Your implementation here
        pass

Coordinator Agent Implementation

Examples

Starter Code