We are hiring an AI Solutions Developer to own that build work end to end.About the Role For each client engagement, we design and build the solution and the data structure behind it, and in most cases we embed AI capability into it. This role owns that work from data model through delivery.There is no single stack. Each client solution gets built on the platform that fits that client: Asana for one, SmartSuite for another, a Zite build for the next, the Microsoft stack for another. Some clients already own a platform and we build inside it. Others need us to choose. You need range across these environments and the judgment to pick correctly, because the platform decision is the hardest one to reverse later.This is not a client-facing role. You take defined specifications from senior leadership, design the data structure, build the solution, layer in AI where the spec calls for it, and hand back something that works, holds up under real use, and can be maintained by someone else.Key Responsibilities Platform selection Assess what the client already owns and what the requirement actually demands, then recommend the platform Justify the choice in writing against the alternatives: cost, data-model limits, permission model, automation ceiling, and AI extensibility When the client already owns a platform, build well inside its constraints rather than fighting it Data architecture Design the data model before building: entities, relationships, keys, constraints, permissions, and how it will grow Apply the same discipline whether the destination is Postgres, SmartSuite, or a set of Asana projects and custom fields Handle migrations and data cleanup without losing or corrupting records Solution build Build the working solution on the chosen platform: intake and record creation, views and queues, role-based access, workflow states, and reporting Build the automation that connects it to the client's existing systems AI integration Build AI capability directly into the solution: document and knowledge retrieval, structured extraction from unstructured inputs, classification and routing, drafting and summarization Build assistants that operate against the client's own data with correct permission boundaries Know when to use a platform's built-in AI and when to build the capability yourself against its API Integration and continuity Connect to CRM, telephony, email, document storage, and ticketing systems through APIs, webhooks, scheduled syncs, and bulk import and export where no native connector exists Fix, extend, and document builds across several different platforms Ensure nothing you build requires you personally to keep it running Requirements Experience Multiple shipped solutions currently in production use that you can walk us through in detail Demonstrable experience carrying a build from a written spec to a working, adopted tool without hand-holding Data modeling (primary requirement) Relational data modeling as a core competency: normalization, keys, cardinality, referential integrity, indexing basics SQL to a working standard: joins across multiple tables, aggregation, subqueries, window functions Experience with at least one of Postgres, MySQL, SQL Server, or Supabase Able to carry a sound data model into a platform that was not built as a database, without ending up with duplicated or orphaned records Recognizes early when a platform's data model cannot support the requirement, and says so before the build starts rather than after Comfortable handling exports of tens of thousands of rows Platform range Deep hands-on build experience in at least three of the following, and the ability to learn a fourth quickly: Asana, SmartSuite, Zite, Airtable, monday.com, ClickUp, Notion, Smartsheet, Microsoft Power Platform with SharePoint and Dataverse Familiarity with app-builder platforms for cases the above cannot carry: Retool, Bubble, Glide, Softr, Budibase, Appsmith, or comparable Automation platforms: n8n, Make, Zapier, Power Automate Able to evaluate a platform honestly against a requirement and say when it is the wrong choice, including when the answer is a platform you have not used before Understands where these platforms break, and when to drop into code rather than force the tool past its limit AI engineering Built production features on LLM APIs such as Anthropic or OpenAI, not prototypes or chat usage Retrieval over client document sets: ingestion, chunking, embeddings, vector storage, relevance tuning Structured extraction: turning unstructured input into validated records that write cleanly into a database Prompt engineering for production: enforced output schemas, failure handling, and no silent hallucination into a data field Accuracy evaluation: how you test that an AI feature is right, and how you catch it when it drifts Token and cost awareness at production volume Sound judgment on where AI belongs and where deterministic logic is the better answer Design sensibility We expect AI to do the heavy lifting on visual output. We do not expect you to be a designer. We do expect taste.You can tell the difference between something that looks credible and something that looks like a free template, and you know which of your outputs is which Working command of visual fundamentals: hierarchy, spacing, alignment, restraint, consistent type and color use Able to direct AI design tools toward a specific intended result and reject what comes back when it is wrong, rather than shipping the first output No decoration for its own sake. Clarity is the standard, and the interface should make the next action obvious to the person using it. Technical foundation REST APIs, authentication patterns, webhooks, pagination, rate limits, and independent reading of API documentation JavaScript or Python for scripting, custom logic, and data transformation Version control, environment separation, and disciplined file and asset organization HTML and CSS at a functional level, sufficient to render output correctly Communication Strong written and spoken English Reads a spec carefully, flags gaps and contradictions before starting, and confirms scope in writing Documents architecture and build decisions as standard practice Preferred Qualifications Has delivered comparable solutions on two or more different platforms and can compare them from real experience Reusable design patterns carried across client accounts and across platforms, rather than rebuilding thinking from scratch each time Platform migration work: moving a client from one system to another with the data and workflows intact Taking over and stabilizing another developer's undocumented build Reporting and analytics layers over operational data Experience working in a client-services model with multiple concurrent accounts Important Fit Note This role is judged on turnaround and reliability, not volume of activity. Scoped requests come back in days, not weeks. You model the data first and confirm the spec before building. You have no favorite tool you push onto every client, and you can learn a new platform in days. You chase down what you need instead of waiting to be unblocked, and your builds handle bad input, edge cases, and volume without babysitting.A candidate who can build competently but cannot explain why they chose a given platform, or who leaves undocumented work behind, is not the right fit for this position.Schedule Full-time.Work Setup Remote. Candidates considered worldwide. Must have a reliable computer, a strong internet connection, and a quiet workspace for calls.
8 days ago