About the Role
I. Role OverviewAs an AI Architect, you are the technical spine of our AI practice. You translate strategic direction set by the AI Advisor into systems that actually work in production - scoping the right architecture, owning the technical design, and ensuring our engineering teams execute with precision and consistency. You are the person who decides how an AI solution gets built, then makes sure it gets built that way.You will work across multiple client engagements simultaneously, partnering closely with an AI Advisor on each - they own the roadmap and the business case; you own the blueprint and the build.II. About YouYou know the difference between an architecture that works in a demo and one that holds up six months after go-live. You can spin up a working prototype in days without cutting corners that matter. You push back on scope that is technically sound but strategically wrong. You are direct with engineers and with clients. And you are genuinely energized by the messiness of applying AI in organizations that weren't built for it. III. Primary ResponsibilitiesDesign end-to-end AI solution architectures: model selection, data pipelines, orchestration layers, integration points, and deployment infrastructureBuild rapid prototypes that make AI concepts tangible for clients - fast enough to drive decisions, rigorous enough to inform the real buildTranslate business requirements and AI strategy into implementation-ready technical specificationsEvaluate and recommend the right components of the AI stack - LLMs, vector databases, fine-tuning approaches, RAG patterns, agents, APIs - based on the client's constraints and goalsDefine and enforce architecture standards across the delivery team; catch design mistakes before they become production problemsLead technical discovery with clients: assess existing data infrastructure, identify gaps, and size the build effort honestlyAct as the senior technical voice in client-facing conversations - not just capable of communicating complexity clearly, but expected to do soReview and guide the work of AI engineers; flag architectural drift early and course-correct without creating bottlenecksContribute to internal IP: reusable patterns, accelerators, and architecture frameworks that raise the floor across all engagementsIV. Ideal Qualifications7+ years in software or data engineering, with at least 3 years in a hands-on architecture roleConsulting or professional services background is required - you understand what it means to deliver under a fixed timeline with a client watchingDeep fluency in designing and deploying production AI/ML systems - not just familiarity with the conceptsPractical experience with LLM application patterns: RAG, agents, function calling, prompt engineering at scale, evaluation frameworksStrong command of at least one major cloud platform (AWS, Azure, or GCP) and its AI/ML services - SageMaker, Azure ML, or Vertex AI - and the ability to architect for cost, latency, and reliability simultaneouslyHands-on experience with LLM orchestration frameworks such as LangChain, LlamaIndex, or similar; vector databases such as Pinecone, Weaviate, or pgvector; and model serving infrastructureProficiency in Python and comfort across the modern data stack: dbt, Airflow or similar orchestration, Snowflake or equivalent cloud data platformsAbility to write and review code - you don't need to be the fastest engineer in the room, but you need to read it fluently and know when something is wrongComfort operating in ambiguous, client-facing environments where requirements evolve and trade-offs are constant.
Originally posted on Himalayas (https://himalayas.app)
