Senior SaaS / AI Engineer

Senior SaaS / AI Engineer — Debug, Repair & Validate Existing Emergent-Built Platform Company: ResFil AI Project Type: Fixed-Price Contract Total Budget: $2,500 Milestones: $500 + $1,000 + $1,000 Platform: Existing Emergent-built SaaS application Scope: Authentication repair, AI integration repair, core feature validation, scheduling/automation testing, production stabilization Important: This is not a rebuild About ResFil AI ResFil AI is an AI-powered customer communication and business automation SaaS platform designed initially for HVAC, plumbing, roofing, and electrical businesses. The existing platform was developed through Emergent and includes functionality such as AI voice receptionists, AI SMS, lead capture, appointment booking, scheduling, knowledge management, customer communication, analytics, follow-up automation, dispatch, customer reactivation, multilingual AI, CRM integrations, and other front-office automation capabilities. We are seeking a senior hands-on SaaS/AI engineer to diagnose, repair, test, and stabilize several critical portions of the existing application. The objective is to repair and validate the current Emergent-generated platform, not rebuild it. Current Critical Issues There are currently two major known issues. Production Login / Authentication Bug Accounts can currently be accessed successfully through the Emergent Preview environment, but login is not working properly through the actual deployed/live ResFil website. We need the engineer to determine why Preview authentication works while Production authentication does not. The finished system must reliably support the applicable account types, including: * Platform administrators * Internal/admin users * Sales/admin representatives * Customer administrators * Customer staff/team users where applicable The engineer should investigate: Authentication → database → production environment variables → cookies/sessions/tokens → domains → API endpoints → CORS → redirects → user records → role permissions → deployment configuration The solution must fix the underlying cause, not create duplicate accounts or temporary workarounds. AI Voice + Chat Business Data Synchronization Business information is successfully entered into the platform, but the AI Voice Agent and AI Chat Agent are not consistently retrieving and using that information correctly. Examples include the AI recognizing a service such as AC Repair while failing to recognize its: * Price * Service area * Appointment duration * Business availability * Scheduling information This requires investigation of the actual runtime path: Database → Backend/API → Tenant Context → AI Tools → Voice/Chat Runtime → Scheduling → Customer Response/Action Milestone 1 — Production Authentication + Technical Diagnosis $500 The first milestone focuses on diagnosis and production access. The engineer will: * Diagnose why authentication works in Emergent Preview but fails on the live deployed site * Repair live-production login * Verify valid sessions are created correctly * Verify appropriate role-based access * Verify admin/customer account login * Test logout * Test invalid credentials * Test expired/invalid sessions * Test protected-route access * Compare Preview and Production environments * Reproduce and diagnose the AI synchronization problem * Identify the exact failing AI/backend layer Milestone 1 Deliverables Provide a concise technical report showing: * Authentication root cause * Production authentication fix * Preview vs. Production differences discovered * Confirmation that relevant admin and customer accounts can log in through the actual live website * AI integration root cause * Affected components * Recommended repair path Milestone 1 must be completed before broader repairs begin. Milestone 2 — AI Integration + Core Product Feature Repair & Testing $1,000 Milestone 2 should repair the central AI data architecture and verify that ResFil’s primary customer-facing features actually work beyond simply appearing in the interface. AI Business Context Repair Both Voice and Chat should dynamically access the correct customer’s: * Knowledge Base * Services * Service descriptions * Pricing * AI price-disclosure settings * Service areas * Appointment types * Appointment durations * Business information * Availability information where applicable The implementation must remain tenant-specific and must not hard-code business information into prompts. Changes made by a business within ResFil should propagate properly to its AI agents. Core Feature Validation The engineer should inspect, test, and repair where necessary the major functionality currently represented as part of ResFil’s subscription offering. This includes, where applicable: * 24/7 inbound AI call answering * AI Voice Receptionist * AI SMS replies * Lead capture * Appointment booking * Call summaries * Call transcripts * Knowledge Base * Internal ResFil scheduling * Basic and enhanced analytics * Dashboard functionality * Customer self-service confirm / reschedule / cancel * Appointment reminders * Calendar integration * Staff/team access limits and permissions * Multiple calendars * Custom/advanced call flows * Advanced scheduling options * Estimate and quote follow-up * Configurable appointment reminders * No-show recovery * Customer reactivation campaigns * Maintenance membership promotion functionality * Seasonal campaigns * Basic technician appointment-status SMS * CRM integrations where currently implemented * Multilingual AI * Automatic language switching * Translated staff transcript functionality * Language analytics The engineer should not assume that a feature works merely because: * A page exists * A button exists * A menu item exists * A database table exists * Demo data appears Each feature should be tested through its actual intended workflow. Example Validation For Lead Capture: Customer interaction → AI identifies lead → lead record created → correct organization → correct customer information → visible within ResFil For Call Summaries: AI call occurs → call is processed → transcript created → summary generated → summary linked to correct customer/call For SMS: Customer sends supported message → AI receives it → correct business context is used → response is sent → conversation is recorded For multilingual functionality: Customer changes language → AI detects/supports language → responds appropriately → relevant transcript/translation information is retained correctly Milestone 2 Deliverables Provide a feature-status report categorizing tested functionality as: * Working * Fixed during milestone * Partially working * Blocked by external configuration * Not yet implemented * Defective / requires additional work Do not claim functionality is working if it was not actually tested. Milestone 3 — Scheduling, Automation, Advanced Features & Full Production Validation $1,000 Milestone 3 focuses heavily on real operational workflows and production readiness. Scheduling & Appointment Booking The AI should be capable of: * Accessing real business availability * Accessing technician/staff availability where applicable * Identifying legitimate appointment times * Respecting existing appointments * Respecting blocked times * Respecting appointment duration * Respecting scheduling rules * Respecting minimum notice * Respecting maximum advance booking * Offering legitimate slots * Booking an actual appointment * Preventing double booking * Displaying appointments within ResFil Scheduling The required workflow is: Caller/message → AI identifies service → AI retrieves availability → AI offers valid time → customer chooses time → appointment created → appointment appears in Scheduling This should be tested through Voice and Chat. Advanced Operational Feature Testing Milestone 3 should also validate and repair where currently implemented: * Advanced scheduling rules * Call transfer rules * Human escalation rules * AI-drafted estimate follow-up * Advanced reminder sequences * Voice + SMS reminders * AI-personalized messaging * Multi-step no-show recovery * Advanced customer reactivation * Outbound AI calling * Custom business workflows * Recurring/automated seasonal campaigns * Smart Dispatch recommendations * Technician ETA / on-the-way notifications * Multi-location routing * Revenue attribution dashboard * Advanced analytics/reporting * Contractor CRM integrations such as ServiceTitan, Housecall Pro, Jobber where currently implemented/configured * Customer self-service portal functionality * Advanced multilingual AI * Language-aware staff routing * Translated staff replies For Custom-plan functionality, validate the underlying architecture where it currently exists, especially: * Custom call flows * Custom scheduling workflows * Custom escalation workflows * Custom business workflows * Custom reminder/communication workflows * Custom no-show logic * High-volume campaign capability * High-volume outbound AI calling * Custom integrations * Multi-location/multi-market routing * Customer portal functionality * Custom attribution/reporting * Enterprise multilingual workflows The engineer is not expected to invent missing Custom customer-specific workflows. The objective is to determine whether the infrastructure that supports those advertised capabilities actually exists and functions correctly. Production End-to-End Validation Before final completion, test the actual live ResFil production environment, not only Emergent Preview. The following chain must work: Live Login → Correct Customer Organization → Business Configuration → Knowledge Base → Services → AI Voice/Chat → Lead Capture → Availability → Scheduling → Booking → Follow-Up/Notifications → Analytics At minimum, test: 1. Admin production login 2. Customer production login 3. Voice agent retrieves business information 4. Chat agent retrieves business information 5. AI correctly states configured services 6. AI correctly states approved pricing 7. AI correctly states configured service area 8. AI correctly states appointment duration 9. AI retrieves legitimate availability 10. AI books appointment 11. Appointment appears in Scheduling 12. Same slot cannot be double-booked 13. Lead is captured correctly 14. Relevant call transcript/summary is generated 15. Customer communications are recorded correctly 16. Applicable reminder/follow-up workflows execute 17. Tenant isolation is maintained Important External Integrations Some functionality may depend on external provider configuration such as: * Twilio * Trust Hub / A2P registration * AI voice provider / Retell * Stripe * Email provider * Google Calendar * CRM providers If something cannot be fully tested because credentials, approvals, or external configuration are not yet available, the engineer should identify it as: BLOCKED BY EXTERNAL CONFIGURATION rather than falsely marking it operational. Experience We’re Looking For We strongly prefer a senior engineer with experience in: * SaaS architecture * AI/LLM tool calling * AI Voice Agents * AI Chat Agents * AI SMS * API integrations * Authentication/session debugging * Production deployment debugging * Multi-tenant SaaS architecture * Scheduling/booking systems * Automation workflows * CRM integrations * Database/backend debugging * Telephony/voice integrations * Existing AI-generated codebases Experience troubleshooting applications where Preview/Staging works but Production fails is particularly valuable. What We Do Not Want We are not looking for someone whose first solution is: “Let’s rebuild the application.” We need someone capable of understanding, debugging, repairing, testing, and stabilizing the existing ResFil platform. We also do not want: * Hard-coded business information * Fake AI responses * UI-only fixes * Features marked working without testing * Temporary authentication workarounds * Production behavior assumed from Preview results Final Deliverable At project completion, provide a concise report containing: Authentication * Preview status * Production status * Admin login PASS/FAIL * Customer login PASS/FAIL AI * Voice business-context retrieval PASS/FAIL * Chat business-context retrieval PASS/FAIL * Knowledge Base PASS/FAIL * Pricing PASS/FAIL * Service Area PASS/FAIL * Appointment duration PASS/FAIL Scheduling * Availability PASS/FAIL * Booking PASS/FAIL * Double-book protection PASS/FAIL * Scheduling record creation PASS/FAIL Core Features * Feature * Plan associated with it * Tested? * Working? * Fixed? * Blocked? * Notes Production * Full production end-to-end test PASS/FAIL * Remaining issues * External blockers * Recommended next steps Budget Milestone 1 — $500 Production authentication repair + technical diagnosis Milestone 2 — $1,000 AI business-data integration + core subscription-feature repair/testing Milestone 3 — $1,000 Scheduling/booking + advanced automation testing + production end-to-end validation Total Fixed Budget: $2,500 When applying, please briefly explain your experience debugging production SaaS systems, AI agents, authentication, LLM tool calling, scheduling systems, API integrations, and multi-tenant applications. Also tell us whether you have previously taken over and repaired an AI-generated or low-code-generated application without rebuilding it from scratch.