Công Ty TNHH Diksha Technologies

Middle Agentic AI Engineer (Enterprise Platforms)

Platforms in Scope

Microsoft Copilot Studio, Google AgentSpace, AWS Bedrock, Dataiku

Role Overview

We are scaling agentic AI across the enterprise using a multi-platform stack including Microsoft Copilot Studio, Google AgentSpace, AWS Bedrock, and Dataiku.

This role designs, builds, and operates production-grade AI agents that can:

• Reason and plan autonomously
• Retrieve enterprise knowledge (RAG)
• Take actions via tools and APIs
• Operate safely under governance and audit constraints

Agents go beyond chat to support real business workflows across IT, operations, security, and enterprise functions.

The role is platform-aware but not platform-locked. Candidates must demonstrate hands-on delivery experience on at least one supported platform and the ability to adapt across others.

What You Will Build

1️⃣ Enterprise Support & Ops Agents

• Learn from tickets, logs, and emails to triage and monitor systems
• Execute or propose actions with escalation and full audit trails
• Perform automated root cause analysis in complex environments

2️⃣ Knowledge & Policy Agents

• Grounded Q&A over SOPs, manuals, and policies
• Citation-enabled, traceable, access-controlled retrieval
• Security attestation and compliance guidance agents
• Onboarding and troubleshooting agents with cost estimation support

3️⃣ Workflow & Action Agents

• Multi-step orchestration (analyze → retrieve → draft → approve → update system)
• API-driven execution with human-in-the-loop controls

Key Responsibilities

🔹 Agentic Solution Design

• Design goal-driven agents with task decomposition, tool selection, state management, and failure recovery
• Implement patterns such as planner–executor, coordinator–worker, reflection/self-check, and human approval gates

🔹 Platform Implementation

Build and deploy agents on one or more of:

• Microsoft Copilot Studio (copilots, plugins, enterprise guardrails)
• Google AgentSpace (workflow orchestration, system integrations)
• AWS Bedrock (secure agentic workflows with models & tools)
• Dataiku (operationalizing agents within ML and analytics pipelines)

Develop hybrid ML + LLM systems:

• ML detects anomalies → LLM explains
• ML predicts incidents → LLM drafts remediation
• ML scores risk → LLM supports decisions

🔹 Retrieval & Grounding (Enterprise RAG)

• Design RAG pipelines: ingestion, chunking, embeddings, retrieval, reranking, citation
• Enforce RBAC and data classification controls

🔹 Evaluation & Operations

• Build evaluation frameworks for non-deterministic systems
• Implement observability across prompts, retrieval, and tool calls
• Own solutions from POC to production

🔹 Security & Responsible AI

• Enforce least-privilege access and audit logging
• Implement prompt-injection defenses and safe action boundaries
• Design approval checkpoints for high-risk actions
• Ensure compliance with enterprise AI governance

Required Qualifications

• 3+ years professional software engineering experience
• Hands-on experience building GenAI / LLM systems beyond basic chatbots (hybrid ML + LLM in production)
• Strong understanding of agentic concepts (tool calling, memory, planning loops)
• Practical experience with at least one: Copilot Studio, AgentSpace, Bedrock, or Dataiku
• Experience with vector databases and embeddings

Preferred Qualifications

• Experience across multiple agent platforms
• Familiarity with enterprise workflows (ITSM, finance, HR, cybersecurity, procurement)
• Experience operating AI under governance and audit constraints
• Knowledge of workflow automation, RPA, or analytics platforms

Benefits & Compensation

• Meal allowance & transportation allowance
• Company-provided laptop
• 100% salary during probation
• Training provided from probation period
• Free access to all LinkedIn Learning courses
• Private health insurance from probation
• Social, Health & Unemployment Insurance (SHUI) paid on full gross base salary
• Annual performance review
• Annual salary review
• Regular company gatherings, team-building activities, and events