Job Description

Senior AI Multi Agent Developer - MINDR Intelligence Stream

About the MINDR Project

Role Overview

The Intelligence Stream serves as the Decide and Act layer of MINDR. As a Senior AI Developer, you will not be designing isolated models or building data pipelines. Instead, you will own the agentic reasoning core, designing and implementing a production-grade Multi-Agent Runtime that orchestrates specialized diagnostic agents, performs automated Root Cause Analysis (RCA), and recommends human-in-the-loop (HITL) remediation actions.

Key Responsibilities

• Design and Develop the Multi-Agent Runtime: Build, optimize, and maintain the deployable Python runtime hosting MINDR agents. You will handle agent lifecycle management, instantiation, state persistence, audit trail logging, and termination.

• Orchestration and Agent Tooling: Build and scale the Principal Agent that dynamically schedules tasks and aggregates results from downstream child agents (Performance, Topology, Alarm, Ticket, CX, and Event Scout) using Model Context Protocol (MCP).

• Implement the RCA Inspector: Program the reasoning engine that consumes multi-source incident context and compares it against structural playbooks (e.g., CDN traffic shift patterns) and historical incidents to produce ranked, explainable root cause hypotheses.

• Construct Stream-to-Stream API Contracts: Maintain tight integration and high API hygiene with neighboring streams: Data Stream (reacting to Anodot anomaly streams via Pub/Sub and consuming Silver layer telemetry), Knowledge Stream (issuing semantic queries to the Knowledge Graph for network topology context), Experience Stream (exposing WebSocket/SSE streams of agent run lifecycle events for real-time operator dashboards).

• Human-in-the-Loop (HITL) Remediation: Build safe, secure integrations for the Remediation Agent to interface with Tardis-fronted APIs (e.g., triggering reroutes, raising CASM tickets, or partner notifications) under strict human approval gates.

Technical Profile Requirement

Must Have Core Skills

• Advanced Python Development: Deep expertise in building scalable production-ready backend services in Python.

• Agentic Orchestration: Hands-on experience developing multi-agent systems, complex prompt engineering, tool use logic, and state management using frameworks or custom runtime loops.

• API Integration Expertise: Proven experience designing clean contract-driven APIs (gRPC, REST, WebSockets, or SSE) and building integrations via the Model Context Protocol (MCP).

• Cloud Stream Infrastructure: Familiarity with the Google Cloud Platform (GCP) ecosystem, specifically Pub/Sub, BigQuery, and Cloud Functions, and streaming architectures (Kafka, CDI).

• Production Diagnostics Mindset: A strong engineering focus on explainability and auditability, ensuring all agent intermediate steps and decisions are logged and reconstructable.

Nice to Have Skills

• Familiarity with Network Topology, IP Core Overload Protection, or telecommunications protocols (VoLTE service chains).

• Experience working with time series anomaly detection engines (e.g., Anodot, Prophet).

• Graph database query experience, semantic queries, graph-based reasoning.

Qualifications

Please Note: Fraudulent job postings/job scams are increasingly common. Beware of misleading advertisements and fraudulent communication issuing 'offer letters' on behalf of T-Systems in exchange for a fee. Please look for an authentic T-Systems email id - XYZ@t-systems.com. Stay vigilant. Protect yourself from recruitment fraud! To know more please visit : [Fraud Alert](Fraud Alert)

Your Task

We are seeking a Senior AI Multi Agent Developer for the MINDR Intelligence Stream project. The Intelligence Stream serves as the Decide and Act layer of MINDR. In this role, you will not be designing isolated models or building data pipelines. Instead, you will own the agentic reasoning core, designing and implementing a production-grade Multi-Agent Runtime that orchestrates specialized diagnostic agents, performs automated Root Cause Analysis (RCA), and recommends human-in-the-loop (HITL) remediation actions.

Key Responsibilities

• Design and Develop the Multi-Agent Runtime: Build, optimize, and maintain the deployable Python runtime hosting MINDR agents. You will handle agent lifecycle management, instantiation, state persistence, audit trail logging, and termination.

• Orchestration and Agent Tooling: Build and scale the Principal Agent that dynamically schedules tasks and aggregates results from downstream child agents (Performance, Topology, Alarm, Ticket, CX, and Event Scout) using Model Context Protocol (MCP).

• Implement the RCA Inspector: Program the reasoning engine that consumes multi-source incident context and compares it against structural playbooks (e.g., CDN traffic shift patterns) and historical incidents to produce ranked, explainable root cause hypotheses.

• Construct Stream-to-Stream API Contracts: Maintain tight integration and high API hygiene with neighboring streams, including the Data Stream (reacting to Anodot anomaly streams via Pub/Sub and consuming Silver layer telemetry), Knowledge Stream (issuing semantic queries to the Knowledge Graph for network topology context), and Experience Stream (exposing WebSocket/SSE streams of agent run lifecycle events for real-time operator dashboards).

• Human-in-the-Loop (HITL) Remediation: Build safe, secure integrations for the Remediation Agent to interface with Tardis-fronted APIs (e.g., triggering reroutes, raising CASM tickets, or partner notifications) under strict human approval gates.

Your Profile

We are seeking a Senior AI Multi Agent Developer with deep expertise in building production-grade multi-agent systems, advanced Python development skills, and strong experience in agentic orchestration and API integration for the MINDR Intelligence Stream project.

Must Have Skills

• Advanced Python development with deep expertise in building scalable production-ready backend services

• Hands-on experience developing multi-agent systems

• Experience with complex prompt engineering

• Experience with tool use logic and state management using frameworks or custom runtime loops

• Proven experience designing clean contract-driven APIs (gRPC, REST, WebSockets, or SSE)

• Experience building integrations via the Model Context Protocol (MCP)

• Familiarity with Google Cloud Platform (GCP) ecosystem, specifically Pub/Sub, BigQuery, and Cloud Functions

• Experience with streaming architectures (Kafka, CDI)

• Strong engineering focus on explainability and auditability

• Experience ensuring all agent intermediate steps and decisions are logged and reconstructable

Nice to Have Skills

• Familiarity with Network Topology

• Knowledge of IP Core Overload Protection

• Experience with telecommunications protocols (VoLTE service chains)

• Experience working with time series anomaly detection engines (e.g., Anodot, Prophet)

• Graph database query experience

• Experience with semantic queries and graph-based reasoning