Engineering Manager - Agentic AI
Telstra · Bengaluru · onsite · Posted today
Undisclosed10–15 yrsfulltimeSourced from
About a minute. You'll make an account on the way through.
About this role
Job Summary
What is the focus of the role
What You ll Do
Agentic AI Solution Engineering
Retrieval and Enterprise Knowledge Systems
MCP, Tools and Agent Communication
Evaluation, Quality and Economics
Security, Governance and Responsible AI
Production Engineering and Operations
Chapter Leadership and Capability Uplift
Job Title: Engineering Manager - Agentic AI
Employment Type: Permanent
What is the focus of the role
The Agentic AI Engineer / Chapter Lead combines hands-on AI engineering depth with chapter leadership. The role designs, builds and operates secure, reliable and cost-effective agentic AI solutions, while developing engineering capability and establishing reusable patterns, standards and reference implementations.
The role works across backend, platform, product, architecture, security and SRE teams to move LLM solutions from prototype to governed production use on AWS, with accountability for technical quality, responsible AI and sustainable adoption of AI-assisted engineering practices.
What You ll Do
Agentic AI Solution Engineering
- Design, build and operate single-agent and multi-agent systems using LangGraph, Amazon Bedrock Agents and appropriate direct SDK approaches.
- Define agent workflows, state, checkpointing, human-in-the-loop controls, tool schemas, contracts, boundaries and safe hand-off patterns.
- Apply engineering judgement to determine when agent frameworks are required and when simpler deterministic or direct model integration is better.
Retrieval and Enterprise Knowledge Systems
- Design and tune RAG and agentic RAG pipelines across approved enterprise knowledge sources, including ingestion, retrieval, indexing, re-ranking and grounding.
- Implement hybrid retrieval, query planning, self-correction and tool-augmented retrieval patterns to improve relevance and task completion.
- Evaluate vector stores and knowledge graph or GraphRAG approaches where scale, security, cost or relationship-heavy domains require them.
MCP, Tools and Agent Communication
- Build and operate MCP servers and clients, including tools, resources, prompts and supported transports.
- Expose secure capabilities with clear descriptions, validation, versioning, least-privilege access and defined trust boundaries.
- Apply agent-to-agent concepts such as discovery, agent cards, task lifecycle, collaboration and capability registration.
Evaluation, Quality and Economics
- Create evaluation frameworks using golden datasets, deterministic checks, LLM-as-judge techniques, human review and regression suites.
- Integrate quality, safety, groundedness, latency, token usage and cost gates into delivery and release processes.
- Benchmark models and architectures for quality, reliability, throughput, context requirements and total cost.
Security, Governance and Responsible AI
- Build security, privacy, safety, guardrails and governance into agentic solutions from inception.
- Mitigate risks including prompt injection, data leakage, excessive agency, unsafe tool use, insecure output handling and supply-chain exposure.
- Implement PII handling, scoped identity, OAuth where applicable, secret management, least-privilege access, human approval and audit trails for sensitive actions.
- Apply data residency, retention, privacy, versioning, hallucination and bias risk assessment, change control and responsible AI requirements.
Production Engineering and Operations
- Build AI services using strong Python practices, including asynchronous programming, FastAPI, automated testing and maintainable package design.
- Deploy and operate services on AWS and Kubernetes with CI/CD, secure runtime configuration and operational controls.
- Instrument agent systems with tracing, logging, metrics and evaluation telemetry using OpenTelemetry and relevant LLM observability platforms.
- Implement reliability patterns including timeouts, retries, fallbacks, failover, rate limits, concurrency controls and graceful degradation.
- Treat prompts, agent definitions, evaluation datasets, policies and configuration as versioned, reviewed and tested engineering artefacts.
Chapter Leadership and Capability Uplift
- Actively build chapter capability by mentoring engineers in software engineering practices, emerging technologies, tools, approaches and delivery processes, modelling continuous learning and coaching others to lift capability across teams.
- Coach, develop and provide regular feedback to chapter members, clarifying performance, behavioural and engineering expectations, and supporting meaningful growth conversations.
- Build strong commercial capability across teams by modelling disciplined cost-value thinking, balancing engineering investment, platform cost, risk and customer outcomes to choose the most economic course of action.
- Foster a culture of observability across multiple teams, helping engineers use operational telemetry, service health data, incident trends and customer-impact signals to improve stability, performance and domain ownership.
- Influence effective decision-making in uncertain or high-pressure situations, making responsible trade-offs without always having complete information and helping teams manage risk, change and ambiguity.
- Build a strong culture of knowledge sharing, documentation and constructive feedback, using communities of practice, playbooks, reference patterns and learning forums to drive continuous improvement and capability uplift.
- Model Telstra s values and facilitate inclusive discussions across teams, creating alignment to organisational strategy and principles while identifying actionable continuous improvement opportunities.
- Manage recruitment of permanent employees and partner resources, shaping a healthy workforce plan, succession pipeline and highly productive team mix aligned to current and future capability needs.
- Set goals and lead performance appraisal for engineers, ensuring expectations, outcomes, development needs and accountability are clearly understood and fairly assessed.
- Ensure co-workers are treated respectfully and without discrimination, harassment or victimisation, actively contributing to psychological safety, inclusion and a respectful workplace.
Sourced from naukri
Skills
BedrockclaudeAI evaluationCrewaiai servicesautomated testingamazon web servicesai engineerChange Controlai securityConcurrencyAsynchronousdeploymentcloudArchitectureAi SolutionsData ScienceArtificial Intelligence
About Telstra
Telstra
Telecom
1.1k+
employees
1901
125 years old
Melbourne, Australia
AU
₹23.4L PA avg
Avg at Telstra
About
Telstra, a leading telecommunications provider in Australia, addresses global connectivity needs with comprehensive communication services and advanced mobile networks.
Industries
Telecom
Employee ratings
10 reviews
3.9/ 5
Culture
3.8
Career growth
3.1
Work-life
4.3
Employees rate it well for
Work-Life Balance43Company Culture40
Find them on