AI Specialist
Description
Responsibilities
Position / Grade : AI Specialist
Vacancy : 1 Position
● Translate business problems from Product Owners and business units into well-framed AI use cases, clarifying root cause, success metrics, and non-AI alternatives before any build decision.
● Build the readiness and impact case for each use case (data availability, governance, risk, human oversight, expected value vs. effort) to support prioritization and go/no-go decisions.
● Author structured, testable specifications (problem statement, scope, acceptance and evaluation criteria) as the primary handoff to AI Engineers within the team's AI delivery workflow.
● Design and run rapid Proofs of Concept using Applied GenAI/LLM techniques, such as prompt design, RAG, tool use, and agentic workflows, to validate feasibility and business value with real users and data.
● Define evaluation methods and quality benchmarks (accuracy, groundedness, safety, latency, cost) and review deliverables against them at each quality gate before promotion.
● Select and justify model, retrieval, and architecture options (hosted LLM APIs, embeddings, vector search; fine-tuning only where justified), documenting trade-offs in cost, performance, security, and vendor lock-in.
● Embed security, privacy, and responsible-AI requirements into every specification and POC in line with company IT security and data-protection policies, and escalate risks early.
● Estimate and monitor token/API usage and cost from POC onward, recommending optimizations so solutions remain economically viable at scale.
● Collaborate with Product Owners, UX/UI, AI Engineers, and IT Security to move validated use cases into production and measure post-launch outcomes against agreed KPIs.
● Codify reusable assets (spec and prompt templates, evaluation datasets, AI skills/agents, playbooks) and share knowledge to raise the AI capability of the DGI team.
● Monitor the evolving AI landscape and bring evidence-based recommendations rather than tool-first adoption.
Qualifications
Must-Have Qualifications
Technical Skills:
● Proven ability to frame business problems as AI use cases and build a readiness and impact case (data, risk, effort, measurable value), including recognizing when a non-AI solution is better.
● Deep hands-on experience with LLMs via APIs and open-weight models, including prompt design, structured outputs, and tool/function calling.
● Hands-on experience designing and building RAG solutions (chunking, embeddings, semantic/hybrid search, vector databases, retrieval quality tuning).
● Hands-on experience building AI agents and agentic workflows, including orchestration, tool integration, and human-in-the-loop controls.
● Ability to design evaluation for LLM systems (test sets, metrics for accuracy, groundedness, and safety, LLM-as-judge with human calibration) and use results to make go/no-go decisions.
● Ability to write clear, structured, testable specifications and acceptance criteria that engineers can build from.
● Strong Python skills for rapid prototyping, data handling, and API integration.
● Working knowledge of LLM cost and performance trade-offs (token usage, model selection, caching, latency).
● Understanding of AI security and responsible-AI practices: prompt injection defense, data privacy, PII handling, and bias/fairness considerations.
Soft Skills:
● Structured problem-solving: separates root cause from symptom and challenges assumptions with evidence.
● Business communication: explains AI options, trade-offs, and risks to non-technical stakeholders and executives in plain language.
● Stakeholder facilitation: runs discovery sessions with Product Owners and business users to reach a clear, agreed problem statement.
● Ownership and judgment: willing to recommend "stop" or "don't build" when evidence does not support the use case.
● Collaboration across Product, UX/UI, Engineering, and IT Security within a defined workflow and quality gates.
● Continuous learning in a fast-moving field, with a critical, evidence-based view of new tools.
Nice to have Qualification:
● Experience with AI-assisted development and coding agents (e.g., Claude Code, Cursor), including building reusable skills or subagents.
● Familiarity with Spec-Driven Development or similar spec-first delivery approaches.
● Experience with Model Context Protocol (MCP) or integrating LLMs with enterprise tools and data sources.
● Experience with red-teaming, adversarial testing, or guardrail/moderation frameworks for LLM applications.
● Experience with Thai-language NLP/LLM use cases.
● Experience with multimodal AI (vision, speech, document understanding).
● Hands-on experience with cloud AI platforms (Google Cloud / Vertex AI, AWS, or Azure) and enterprise data platforms (e.g., Snowflake).
● Domain exposure to automotive, connected vehicles, dealer/sales operations, or loyalty programs.
● Product management or design-thinking background (e.g., Agile, Jira, discovery methods).
● Contributions to open-source AI projects or community knowledge sharing.
Skills
Sign in to read and join the discussion.