Chi tiết công việc AI Agent Engineer tại bTaskee Vietnam
Role Overview:
We are looking for a Middle/Senior AI Agent Engineer to design, build, and harden autonomous and semi-autonomous AI agents that operate within bTaskee's product ecosystem (customer chat support, Tasker assistance tools, internal operations automation, and more). A core focus of this role is harness engineering – building the evaluation, testing, and orchestration infrastructure that let us safely develop, test, and continuously improve agent behavior before and after it reaches production.
You'll work closely with product, backend, and data teams to turn LLM capabilities into reliable, measurable, and maintainable agent systems – not just prompts that work once in a demo.
Key Responsibilities:
- Design, develop, and maintain multi-agent systems and autonomous AI Agents (tool-use, multi-step reasoning, memory/state management) for customer-facing and internal bTaskee products.
- Build and maintain agent harnesses: test environments, simulated user/task scenarios, scoring rubrics, and regression suites that evaluate agent behavior across many conditions before deployment.
- Define and track evaluation metrics (task success rate, hallucination rate, latency, cost per interaction, tool-call accuracy) and build dashboards/pipelines to monitor them continuously.
- Integrate agents with internal APIs, databases, and third-party tools (search, payments, scheduling, notifications) via structured tool-calling.
- Own the full agent lifecycle: prompt/context design, tool orchestration, guardrails, fallback logic, and production monitoring.
- Run systematic experiments (prompt variants, model versions, retrieval strategies) using the harness to compare performance objectively rather than by eyeballing outputs.
- Stay current on agent frameworks, evaluation methodologies, and LLM tooling, and bring pragmatic recommendations back to the team.
Qualifications and Skills:
- 3+ years of hands-on experience building AI Agent/LLM-powered applications;
- Hands-on experience building or operating agent harnesses/eval frameworks – i.e., systems for automated testing, scoring, and regression-checking of AI agent behavior (not just manual prompt tweaking);
- Strong proficiency in Python (or similar) and experience with agent/orchestration frameworks (e.g., LangChain, LangGraph, LlamaIndex, Anthropic/OpenAI agent SDKs, or custom-built equivalents);
- Practical experience with LLM tool-calling/function-calling, structured outputs, and context/memory management;
- Experience with production ML/AI systems: logging, tracing, monitoring, and incident debugging;
- Comfortable working with REST/GraphQL APIs, databases, and cloud infrastructure (AWS/GCP);
- Solid understanding of evaluation methodology for generative AI systems: designing test sets, scoring rubrics, human-in-the-loop review, and statistical comparison of results.
Nice to have:
- Experience with RAG pipelines and vector databases.
- Experience in on-demand/marketplace or customer-support automation products.
- Prior work on multi-agent systems or agent-to-agent coordination.
At bTaskee, we don’t just work – we thrive with top-notch benefits!
- Competitive Salary – Rewarded based on your experience and skills;
- Annual Performance Review – Unlock growth opportunities every year;
- KPI-Based Salary – Earn what you deserve through your contributions;
- 13th-Month Bonus – Rewarded based on both your performance and the company’s success;
- Equipment Provided – Enjoy top-tier equipment that enhances productivity and makes your daily work easier;
- BYOD Support – Get an allowance while enjoying the flexibility of using your own laptop;
- Comprehensive Healthcare – Regular check-ups & a premium health package;
- SHUI Compliance – Contributions aligned with legal regulations;
- Generous Leave Policy – Enjoy 12–16 annual leave days per year;
- Celebration Perks – Special benefits for birthdays, weddings, childbirth, and more;
- Engaging Workplace – Activities & exciting team-building events;
- Advance Home Care Benefit – Get a monthly bTaskee package, freeing up your time for learning, relaxation, and self-care.
