GEN AI Solution Architect
Artech Information Systems View all jobs
- Toronto, ON
- Contract
- Full-time
Location: Toronto - Hybrid (3-4 days WFO)
Duration: 6 MonthsPay range: C$55 incKey Responsibilities:Product Roadmap
Modular Design
- Define the product vision and roadmap for reusable Gen AI modules (e.g., RAG, prompting frameworks, hybrid MLLLM systems).
- Architect parameterized, business-agnostic solutions that abstract complexity (e.g., pre-configured prompts, vector DB connectors, chunking logic)Design APIs and microservices to expose modules as reusable components (e.g., text-to-SQL service, RAG-as-a-service).
Technical Leadership
- Standardize patterns (e.g., prompt templates, chunking strategies, few-shot training pipelines) across use cases
- Integrate LLM workflows (e.g., OpenAI, Claude) with traditional ML (clustering, classification) and enterprise systems (databases, UI tools).
- Optimize performance of Gen AI components (cost, latency, accuracy) and ensure scalability (e.g., load balancing for vector DBs).Adoption Enablement
- Develop documentation, tutorials, and sandbox environments for testing modules.
- Train teams on best practices (e.g., prompt engineering, security for LLM outputs)Track metrics
Module reuse rate*** contribution volume, time-to-deploy for new use cases.Required SkillsTechnical Expertise
Gen AIML Engineering - Hands-on experience with LLM integration (e.g., OpenAI, Anthropic, Llama 2) and frameworks (LangChain, LlamaIndex).
- Expertise in RAG workflows Document chunking (sentence transformers), vector DBs (Pinecone, FAISS), and hybrid search
Familiarity with text-to-SQL systems, few-shotchain-of-thought prompting, and traditional ML (clustering with scikit-learn, PyTorch).
- Software Engineering - Proficiency in Python, API design (FastAPI, Flask), and cloud platforms (AWS Sagemaker, Azure AI).
- Experience with CICD, containerization (Docker), and infrastructure-as-code (Terraform).UI Integration Skills
- Frontend integration (ReactStreamlit for config UIs) and middleware (message queues, auth systems like R2D2).Product Strategy
- Proven track record of building reusable MLAPI products or internal platforms.
Comments for Suppliers: