Tweet
Facebook
Copy Link
Mô tả công việc
The core product experience runs on an LLM- driven agent that extracts data from client- submitted receipts and documents over WhatsApp, proposes accounting entries, and answers staff questions inside the web app. We’re hiring an AI/Agent Engineer to take ownership of this entire layer, making it cheaper, safer and more reliable while closing specific security gaps (prompt injection, unchecked tool authorization) identified in a recent technical audit. You will work alongside a backend- leaning Fullstack Engineer, who owns the surrounding application and infrastructure.
ResponsibilitiesReplace the agent’s currently hard- coded (effectively disabled) permission model with real, enforced per- tool authorization checks
Evaluate and implement model tiering, cheaper models for extraction, stronger models for judgment calls and plan a fallback provider for resilience
Advise on data- processing- agreement and data- residency questions for the AI sub- processors (Anthropic, OpenAI), alongside the company’s data- protection advisor
Implement PII redaction before data is sent to third- party LLM providers, and partner with the Fullstack Engineer to log every agent tool call to an audit trail
Own the LLM integration layer used for document extraction (OCR), accounting- entry proposals, and the in- app chat agent, currently built on Anthropic/OpenAI APIs
Harden the agent against prompt injection: treat all extracted document text as untrusted input, and require explicit confirmation gates before any tool call that grants data access, creates a contact, or moves money
Build an evaluation/golden- file test harness so model or prompt changes can be validated before shipping to production
Implement prompt caching to cut redundant token usage. An identified ~83% cost- reduction opportunity and introduce per- tenant usage budgets and quotas
Contribute to the n8n- side automation where it touches AI extraction, the document- intake pipeline and coordinating handoffs with the Fullstack Engineer.
Yêu cầu công việc
Yêu cầu công việc
Working knowledge of security concerns specific to LLM systems: prompt injection, data exfiltration via generated content, unsafe tool authorization
Experience building or using evaluation frameworks/golden- file testing for LLM- based systems
Hands- on experience with agentic tool- use patterns, function/tool calling, multi- step agent loops, not just single- shot prompting
Clear written communication in English; able to work asynchronously across time zones
3+ years of experience building production applications on top of LLM APIs (Anthropic Claude, OpenAI, or equivalent)
Practical understanding of LLM cost structures and optimisation techniques, prompt caching, context trimming, model tiering
Strong prompt- engineering skills, including context management and techniques for defending against prompt injection
Proficiency in Python or TypeScript, sufficient to build and modify a backend AI- integration layer
Quyền lợi
You’ll love working here!
Mon- Fri, flexible remote working.
English, professional working environment.
Competitive salary package.
Opportunities to gain hands- on experience in cutting- edge technologies.
Premium health care package for employees & family members.
Annual leave increases with loyalty.
12 days of annual leave and 3 additional special holidays (Swedish National Day, Christmas Eve & your Birthday).
Cập nhật gần nhất lúc: 2026-09-16 11:45:04
fram^ là một công ty tư vấn và phát triển CNTT trực tuyến có văn phòng tại Thụy Điển (Stockholm) và Việt Nam (Thành phố Hồ Chí Minh). Chúng tôi có niềm đam mê cháy bỏng trong việc phát triển các giải pháp CNTT trực tuyến tuyệt vời với một đội ngũ sáng lập đã xây dựng và quản lý các doanh nghiệp trực tuyến hàng đầu thị trường tại Thụy Điển và Đông Nam Á. Chúng tôi giúp các công ty ở Scandinavia và Bắc Âu phát triển các giải pháp CNTT trực tuyến đẳng cấp thế giới theo cách chuyên nghiệp và tiết kiệm chi phí.
Tính năng đăng nội dung chưa được mở
Note8.vn đang hoàn thiện thủ tục cấp phép mạng xã hội. Trong thời gian này, bạn vẫn có thể xem nội dung và sử dụng các tiện ích khác.