Sr. Applied AI Engineer

Location: Remote (Nepal)
ABOUT THE ROLE
We are looking for a senior applied AI engineer to join our team in Nepal. You will be assigned to a client project and you'll build production-grade AI agents that automate real business operations. They'll sit on top of an application and API layer our team already built and maintains, and they'll integrate with third-party systems.
This is a hands-on engineering role, not a research or architecture-only position. You'll work directly with the client's operations stakeholders to understand how work actually gets done, decide what should and shouldn't be automated, and ship reliable systems into production.
You'll work closely with our existing backend and frontend engineers. They own the core application; you'll own the agent layer, and you'll collaborate on the APIs, tools and human-approval interfaces the agents need.
WHAT YOU'LL DO
- Work with client stakeholders to map business workflows and identify where AI agents add real value and where deterministic logic is the better choice.
- Design and build agents that carry out multi-step workflows using tool calling, structured outputs, retrieval, state management and human approval.
- Expose existing backend capabilities as well-defined agent tools, working with our engineering team.
- Integrate agents safely with third-party APIs and business systems.
- Build evaluation pipelines using representative historical cases to measure agent accuracy, failure modes, latency and cost, and use them to drive improvements.
- Implement least-privilege tool access, authorization, audit logging, prompt-injection mitigation and guardrails around every action an agent can take.
- Design for failure: retries, timeouts, idempotency, partial execution, malformed model outputs and third-party outages.
- Set up observability so any agent run can be traced, replayed and debugged.
- Deploy and operate agents in production, and manage token costs through model selection, routing, caching and context management.
- Build reusable internal patterns, libraries and practices that let us deliver agentic systems for future clients faster.
- Evaluate new models and tools, and adopt them when they bring a measurable improvement.
WHAT WE'RE LOOKING FOR
Must Have
- 4+ years of professional software engineering experience.
- Strong proficiency in Python, TypeScript or Node.js.
- Solid backend fundamentals: REST APIs, authentication and authorization, databases, asynchronous processing and queues.
- Hands-on experience building applications on modern LLM APIs, beyond simple prompt-and-response usage.
- Experience building multi-step LLM workflows or agents that use tools, APIs or external systems.
- Experience evaluating LLM systems systematically, for example with test sets, regression evals or failure categorization, rather than relying on manual spot-checks.
- A working understanding of AI security risks, including prompt injection, excessive agent permissions, data leakage and PII handling, and how to mitigate them.
- Experience deploying and operating software in production, including testing, monitoring and incident debugging.
- Ability to design reliable systems around probabilistic outputs.
- Clear communication skills and comfort working directly with non-technical stakeholders to translate business processes into technical designs.
Preferred
- Priority will be given to candidates with 8+ years of software engineering experience, particularly in backend and SaaS systems. We value deep production engineering judgment applied to AI.
- Experience with agent frameworks or SDKs such as LangGraph, provider agent SDKs, or similar. We care more about understanding agent primitives than about any specific framework.
- Experience with MCP or similar approaches for connecting agents to tools and systems.
- Experience with RAG, embeddings, vector databases or hybrid search.
- Experience with LLM observability and evaluation tools.
- Experience with AWS, Azure or Google Cloud, including managed LLM offerings.
- Experience integrating with business platforms such as CRM, ERP, support, communication or financial systems.
- Experience in a consulting, agency or client-facing engineering role.
Not Required
- A PhD or research background.
- Experience with model training, fine-tuning, or frameworks like PyTorch or TensorFlow.
- Computer vision or NLP research experience.
WHAT WE VALUE
Not every problem needs an AI agent. We want engineers who can confidently decide when to use an LLM, when to use deterministic logic, and how to combine the two into systems that are reliable, maintainable and cost-effective.
We value engineers who think past the model to authorization, audit trails, failure recovery, cost and the everyday reality of production software. Someone who asks "what happens when this goes wrong?" before "which framework should we use?"
THE IDEAL CANDIDATE
You're a seasoned software engineer who has moved seriously into applied AI. You can take a workflow like:
Business process → context → retrieval → reasoning → tool calls → decisions → action → escalation when needed
and turn it into a production system that's observable, auditable and measurably accurate.
You don't need to be an AI researcher. You do need to be an excellent engineer who understands what modern AI can and cannot reliably do.
COMPENSATION
Negotiable / Commensurate with experience and skills
EDUCATION
A degree in Computer Science, Engineering or a related field is helpful, but equivalent professional experience and demonstrated ability are equally valued.

Work arrangement
Yes