AI Engineer at The Stakeholder Company Pte. Ltd.

Ho Chi Minh City, , Vietnam -

Full Time

Start Date

Immediate

Expiry Date

02 Oct, 26

Salary

0.0

Posted On

04 Jul, 26

Experience

2 year(s) or above

Remote Job

Yes

Telecommute

Yes

Sponsor Visa

Skills

Python, SQL, LLM Applications, Retrieval-Augmented Generation, Machine Learning, Data Pipelines, API Design, Cloud Computing, Production Observability, CI/CD, Prompt Engineering, Entity Resolution, Model Fine-tuning, Automated Testing, Version Control, System Design

Industry

technology;Information and Media

Description

Vietnam | Hybrid, Ho Chi Minh City | Reports to: Head of AI Role Overview TSC builds AI-native stakeholder and issue intelligence products for enterprise teams operating across public affairs, regulatory, reputational and external-risk environments. We are hiring an AI Engineer to build and operate the AI systems behind Genie, turning fragmented, fast-changing data into reliable and actionable intelligence. This is a product-focused engineering role. You will move from rapid experimentation to production delivery, working across LLM applications, retrieval, machine learning, data pipelines, APIs, evaluation and observability. You will work closely with the Head of AI, Product, Data Engineering and Platform Engineering across Vietnam and Singapore. Success is not measured by the novelty of a model or the sophistication of a demonstration. It is measured by whether the system produces accurate, traceable and useful outputs for customers, operates safely, and performs at sustainable cost and latency. The Mandate Build production AI capabilities that convert noisy public, proprietary and customer data into dependable stakeholder and issue intelligence. What You Will Own 1. Production AI Systems Design, build, test and operate AI services and workflows from proof of concept through production deployment. Develop capabilities across classification, extraction, summarisation, entity resolution, stakeholder mapping, retrieval, risk detection and insight generation. Select the appropriate combination of deterministic software, machine learning, retrieval, LLMs, agents and human review for each use case. Build clear APIs and integrations that connect AI capabilities to Genie and related internal systems. Own the quality, reliability and maintainability of the components you ship, including production support and continuous improvement. 2. Data, Retrieval and Model Quality Build and improve pipelines for ingestion, cleaning, enrichment, labelling, chunking, indexing and retrieval. Improve retrieval relevance, grounding, entity resolution and freshness across fragmented real-world datasets. Fine-tune or adapt specialised and open-source models where this provides a clear quality, privacy, latency or cost advantage. Create reusable datasets, prompts, configurations and evaluation artefacts rather than relying on one-off experiments. Partner with Data Engineers and domain experts to turn feedback and edge cases into systematic quality improvements. 3. Evaluation, Reliability and Safety Define and implement automated and human evaluation for AI features, including golden datasets, regression tests and task-specific quality metrics. Instrument systems with logs, traces, latency and token monitoring, quality signals and actionable failure diagnostics. Design appropriate fallbacks, retry behaviour, human-review thresholds, rollback paths and incident escalation. Protect sensitive and customer-specific data through secure handling, access controls, workspace boundaries and careful use of external model providers. Identify and mitigate risks including prompt injection, data leakage, stale context, unsupported claims, excessive autonomy and unsafe tool permissions. 4. Product Delivery and Technical Judgement Translate ambiguous customer and business workflows into clear technical requirements, experiments and acceptance criteria. Ask who will use an output, what decision it supports, what failure would cost, and how its value can be measured. Work with Product to sequence experiments and releases based on customer value, quality, feasibility and operational risk. Communicate technical trade-offs, limitations and evidence clearly to both technical and non-technical colleagues. Contribute to architecture and roadmap decisions through working software, evaluation evidence and documented recommendations. 5. Agentic and AI-Assisted Engineering Build bounded agentic workflows with explicit tool permissions, validation steps, stopping conditions, budgets and human approval points. Use coding assistants and engineering agents to improve delivery speed while remaining accountable for design, security, testing and code quality. Prefer deterministic orchestration when it is safer, cheaper or easier to test than an autonomous agent. Evaluate emerging tools through controlled experiments and adopt them only when they improve measurable product or engineering outcomes. 6. Global Engineering Collaboration Work as a core member of TSC's global engineering team, with regular collaboration across Vietnam and Singapore. Maintain clear technical documentation, implementation notes and decision records in shared systems such as Jira and Confluence. Participate actively in code reviews, design reviews, sprint planning, incident learning and architectural discussions. Use overlapping working hours for high-value collaboration while preserving focused execution through strong asynchronous communication. What Success Looks Like Within the first 6 months, you will have: Shipped and operated material AI capabilities used in live customer or internal workflows. Established evaluation and observability for the features you own, with clear quality baselines and regression controls. Improved at least one critical dimension of a production workflow, such as accuracy, retrieval quality, latency, cost, traceability or human-review effort. Strengthened reusable data, retrieval and model components rather than accumulating isolated prototypes. Demonstrated reliable ownership across design, implementation, documentation, deployment and production support. Built trusted working relationships with Product, and Engineering across Vietnam and Singapore. Requirements Must-Have Typically 4+ years of experience in software, data, machine learning or AI engineering, with evidence of shipping and operating production systems. Stronger direct experience may substitute for a specific tenure threshold. Professional proficiency in Python and practical experience building tested, maintainable services or data workflows. Strong SQL capability is expected. Hands-on experience with LLM applications, retrieval-augmented generation, embeddings, structured extraction, orchestration and evaluation. Experience with data pipelines, APIs, version control, automated testing, CI/CD and cloud environments such as GCP or AWS. Working knowledge of production observability, including logging, tracing, latency, failures, model usage and cost monitoring. Sound judgement around sensitive data, model limitations, prompt injection, tool permissions, human review and rollback. Professional English communication skills, particularly for technical documentation, code review and cross-border architectural discussion. A bachelor's degree in computer science, engineering or a related field, or equivalent practical experience. Strong Signals You have built AI products that use classification, named-entity recognition, entity resolution, summarisation, extraction or recommendation over messy real-world data. You have created evaluation datasets or automated quality checks for LLM, RAG or machine-learning systems. You have used orchestration frameworks such as LangGraph or LangChain, while remaining able to design systems independently of a specific framework. You have deployed or fine-tuned open-source models and understand the trade-offs against managed model APIs. You have improved production model quality, latency or cost through routing, caching, batching, context management or model selection. You have worked in a multi-tenant enterprise environment or with confidential customer data. You can show production work, open-source contributions, technical writing or well-explained experiments that demonstrate how you solve engineering problems. This Role Is Not A research-only role focused primarily on novel models, publications or benchmark performance. A prompt-engineering role that stops at demonstrations and does not own code, data, evaluation or production reliability. A framework-chasing role where new tools are adopted without evidence of customer or engineering value. A narrow modelling role isolated from APIs, data pipelines, deployment and the wider product lifecycle. Why Join TSC Build AI systems for real enterprise decisions, not speculative demonstrations. Work directly with the Head of AI, Product and senior engineering leaders across Vietnam and Singapore. Help shape TSC's AI engineering practices and the continued development of its Vietnam engineering capability. Work in a hybrid model that combines office collaboration in Ho Chi Minh City with focused remote execution. Receive a competitive local package, equity options and a 13th-month bonus. Develop across applied AI, data engineering, product judgement and enterprise-grade delivery. The Core Question Can you turn an ambiguous intelligence problem and messy data into a production system that users can trust, engineers can operate, and the business can scale?

Responsibilities

Design, build, and operate production AI systems and workflows to convert noisy data into actionable stakeholder intelligence. Own the end-to-end lifecycle including data pipelines, model evaluation, reliability, and integration with internal APIs.