AI Browser Automation Specialist

We're looking for a specialist to extend our existing browser extension with intelligent auto form-fill capabilities. The feature fills lender application forms on third-party websites using data from our platform — automatically, reliably, and fast. This role sits at the intersection of browser extension development, DOM engineering, and applied AI. You'll be solving real-world messiness: forms that change without notice, validation rules that differ from what our data provides, and multi-step flows that need to navigate themselves. What You'll BuildWorking within our existing extension codebase, you'll implement a form-fill feature that: Reads structured applicant data already captured by the extension from our platform Automatically fills multi-step lender application forms (e.g. credit application portals) Navigates between form steps autonomously Detects and self-corrects validation errors — including fuzzy data mismatches between our platform data and what the form expects (e.g. date formats, field length constraints, dropdown option matching) Self-heals when form structure changes — using AI to adapt to minor DOM shifts (renamed fields, reordered steps, changed selectors) without code changes. Significant structural overhauls (full page redesigns, new multi-step flows) may require a targeted code update, but the goal is to minimise how often that's needed. Prioritises speed: static DOM injection for known fields, AI only where it adds value (error correction, unknown fields, structural changes) Technical Requirements Solid experience contributing to existing browser extension codebases (Chrome/Manifest V3) Deep knowledge of DOM manipulation, form interaction, and event simulation — including framework-controlled fields (React, Vue, Angular) Able to design hybrid pipelines: fast static fill for the happy path, AI fallback for edge cases Practical experience integrating LLM APIs (OpenAI, Anthropic, or similar) into production features Understanding of form validation patterns and how to detect and respond to them programmatically Nice to Have Experience with self-healing selector strategies (semantic fallbacks, selector scoring, visual anchoring) Prior work on form automation across multiple third-party portals (fintech, insurance, automotive lending) Familiarity with prompt engineering for structured data correction tasks Experience working in codebases that mix static automation with AI-assisted fallback logic What Success Looks Like Form fill completes end-to-end without human intervention on the majority of submissions AI-assisted error correction handles common data mismatches automatically When a lender updates their form, the system degrades gracefully and self-corrects rather than silently failing Fill speed is not noticeably impacted by AI calls — AI runs only on failure paths or where static fill isn't sufficient Engagement [Contract / Full-time — adjust as needed] Opportunity for full time role applies. Remote-friendly. Must be able to collaborate async with our engineering team.