ABOUT THIS JOB
- OmniPhi is building the first Integrated Agentic Trading Environment (IATE), bringing strategy research, development, testing, deployment, and live operation into one AI-native platform.
- We are hiring a Senior Agentic AI Engineer to improve our core agentic systems, including the agent harness, tool use, memory, context management, RAG, multi-agent coordination, reliability, and efficiency.
- This is a hands-on engineering role requiring experience building and improving production agentic systems beyond basic chatbots and API wrappers.
RESPONSIBILITIES
- Design, build, and continuously improve the core agent harness.
- Develop reliable tool creation and tool-calling systems across internal tools, external APIs, databases, and MCP connections.
- Implement structured outputs, tool schemas, validation, permissions, retries, and failure recovery.
- Build persistent sessions, agent memory, state management, delegation, and multi-agent workflows.
- Develop RAG systems using ingestion pipelines, embeddings, vector and hybrid search, reranking, and citations.
- Optimize prompts, context assembly, token efficiency, caching, model routing, latency, and inference costs.
- Build evaluations, tracing, monitoring, guardrails, and regression testing for agent behavior and tool execution.
- Integrate commercial and open-weight language models into reliable production services.
REQUIRED QUALIFICATIONS
- Bachelor’s degree in Computer Science, Artificial Intelligence, Machine Learning, Software Engineering, or a related field, or equivalent practical experience.
- 5+ years of software engineering, machine learning, or AI engineering experience.
- Production experience building agent harnesses, agent runtimes, or complex LLM applications.
- Advanced Python skills, including asynchronous programming, APIs, testing, and production debugging.
- Strong Rust skills, with the ability to build, maintain, and contribute to production Rust services.
- Strong understanding of prompt engineering, context engineering, tool calling, MCP, and structured outputs.
- Experience with RAG, agent memory, embeddings, vector databases, and retrieval optimization.
- Experience with LangChain, LangGraph, LlamaIndex, PydanticAI, or comparable technologies.
- Experience evaluating and improving agent quality, reliability, latency, token usage, and cost.
STRONGLY PREFERRED
- Familiarity with PyTorch, TensorFlow, Hugging Face, ONNX, or vLLM.
- Fine-tuning experience using LoRA, QLoRA, PEFT, or comparable techniques.
- Knowledge of quantization, distillation, and efficient model inference.
- Understanding of mixture-of-experts architectures and model-routing systems.
- Knowledge of machine learning, deep learning, and natural language processing fundamentals.
- Experience with SQL, PostgreSQL/pgvector, Redis, data pipelines, cloud services, and Linux.
- Experience with coding agents, sandboxed execution, or durable workflow systems.
- Familiarity with TypeScript, Java, or Bash is an asset.
WHAT WE’RE LOOKING FOR
- You do not need experience with every framework or technology listed above.
- We are looking for someone who understands the underlying architecture of modern agentic systems.
- You should have personally built or made substantial contributions to meaningful production agent systems.
- You should be able to clearly explain your architectural choices, technical decisions, and trade-offs.
Work Location: Hybrid remote in Vancouver, BC