Job location:: Toronto, Brampton, Canada

About the Role

You will implement the mechanical core of our AI features: chunking logic, embedding flows, retrieval algorithms, agent state machines, and prompt-construction engines.

You will ensure every component is observable, testable, and optimized for enterprise workloads.

What You Will Do

- Build RAG Pipelines: Implement custom chunking, embeddings, hybrid search, re-ranking, and retrieval logic tailored to domain-specific semantics.
- Agentic Orchestration: Build multi-step agents with working memory, tool execution, state tracking, and deterministic control flows.
- Context Engineering: Optimize prompts, context packing, and token-economics to maximize reasoning quality while minimizing latency and cost.

Required Qualifications

- Strong software engineering fundamentals with intermediate Python.
- Experience building transparent AI systems using standard libraries and HTTP clients.
- Knowledge of Google GECX.
- Experience with structured extraction using JSON schemas, Pydantic,



and advanced prompting

Skillset Requirements

- Native RAG Implementation: Custom chunking, embeddings, hybrid search, re-ranking, and retrieval logic.
- Agentic Programming: Building tool-use flows, working memory, state machines, and deterministic agent orchestration.
- Prompt Engineering: Crafting structured prompts, multi-shot reasoning scaffolds, and domain-specific
- Python Engineering: Solid fundamentals, async programming, concurrency, and performance tuning.
- Structured Extraction: JSON schema design, Pydantic models, and deterministic extraction patterns.
- LLM Tooling: Experience with HTTP clients, raw API calls, and minimal-framework AI development.
- Testing & Debugging: Unit tests for agents, RAG regression tests, and prompt-level debugging.
- Observability: Instrumenting tracing, logging, and token/latency metrics for agents and RAG components

#J-18808-Ljbffr

📌 AI Developer (Toronto)
🏢 Ampstek
📍 Toronto