Posted on 2026/10/07
AI Systems Engineer. Agentic Automation & Local LLMs
BricksFolios Real Estate Solutions
Pakistan
BricksFolios | Full-Time | Remote
We are a U.S.-based wealth-tech platform helping property owners and real estate investors make smarter decisions across leasing, property management, portfolio strategy, and long-term wealth creation.
Our founders have held senior leadership roles at companies including Microsoft, Accenture, GE, and EY, and built BricksFolios after experiencing firsthand how fragmented and inefficient the traditional real estate industry can be.
Today, we are combining technology, AI, real estate operations, and investment thinking to build a more intelligent way to own and manage residential real estate.
Specifically, we're building a 24/7 agentic operating system across marketing, sales, leasing, property management, short-term rentals, and customer operations.
Our goal is straightforward:
Automate everything that should be automated.
Keep humans involved where judgment, risk, or relationships matter.
We are looking for a hands-on engineer who can take a real business workflow, understand how it works, architect the right solution, connect multiple systems, write code where needed, deploy agents, and continuously improve them.
This is not a ChatGPT + n8n workflow-builder role.
We want someone who understands AI systems deeply enough to know when to use an LLM, when to use a smaller local model, and when ordinary software is the better answer.
Marketing & Lead Generation
Build agentic workflows for:
- Prospect discovery and list building
- Data enrichment and intent signals
- Lead scoring and segmentation
- Personalized email/SMS outreach
- Campaign execution and monitoring
- Response classification
- Lead nurturing and reactivation
- Content and campaign automation
- Attribution and CRM synchronization
Sales & CRM:
Turn our CRM into an actively managed, intelligent pipeline:
- Lead qualification
- Automated follow-up
- Appointment scheduling
- Conversation intelligence
- Pipeline and stage management
- Stalled-lead identification
- Next-best-action workflows
- Human escalation for high-intent opportunities
Leasing:
Automate the journey from rental inquiry to signed lease:
- Inquiry response
- Prospect qualification
- Self-tour scheduling
- Tour reminders
- Post-tour follow-up
- Objection identification
- Application nudges
- Lead routing
- Leasing team escalation
Property Management:
Build 24/7 agents capable of supporting:
- Tenant communications
- Maintenance intake and triage
- Emergency identification and escalation
- Vendor dispatch and coordination
- Work-order follow-up
- Rent and delinquency workflows
- Lease renewals
- Inspections
- Owner communications
- Document processing
- Exception management
The objective isn't simply to create a maintenance ticket.
An agent should be capable of understanding an issue, gathering context, determining the appropriate workflow, communicating with the parties involved, tracking the issue through resolution, updating our systems, and involving a human when necessary.
Short-Term Rentals & Operations:
Automate:
- Guest communications
- Booking workflows
- Check-in/check-out
- Issue triage
- Cleaning and turnover coordination
- Operational alerts
- Reviews
- Reporting
- Repetitive back-office processes
Our Technology Environment
You'll build across our existing systems:
- Follow Up Boss. CRM and lead management
- ShowMojo. listing syndication and self-tour management
- Rentvine. property management
- Twilio. communications
- n8n. workflow orchestration
- BricksFolios proprietary platform. Node.js, AngularJS, MongoDB and AWS RDS
- Local/open-source LLMs
- Frontier LLMs where their capabilities justify their use
- APIs, webhooks, databases and event-driven integrations
We don't want hundreds of disconnected automations.
We want an intelligent operating layer across these systems that understands context, maintains state, determines what should happen next, takes approved actions, and escalates exceptions.
How We Want You to Think
A typical AI automation might look like:
Trigger → Send everything to GPT → Response → Next node
That's not what we're trying to build.
We want engineers who think more like:
Event → State → Deterministic Logic → Retrieval → Model if Needed → Action → Validation → State Update → Next Action
Before calling an LLM, you should naturally ask:
Does this actually require an LLM?
We care about:
- Token-efficient architecture
- Local LLM deployment
- Model routing
- Structured state
- Context engineering
- RAG where appropriate
- Deterministic workflows
- Tool/function calling
- Reliability and failure recovery
- Human-in-the-loop controls
- Observability
Sometimes the best AI architecture uses less AI.
What You'll Own
You'll take workflows from:
Business Problem → Architecture → Build → Deploy → Monitor → Improve
Responsibilities include:
- Map business workflows and identify automation opportunities
- Architect multi-step agentic systems
- Build agents and backend services
- Deploy local/open-source models where appropriate
- Integrate third-party APIs and webhooks
- Build directly into our proprietary platform when appropriate
- Design state and memory architecture
- Build RAG and knowledge-retrieval systems
- Implement tool calling and structured outputs
- Design retries, fallbacks and human escalation
- Build monitoring and logging
- Diagnose production failures
- Optimize model and token usage
- Continuously identify additional work that can be automated
Technical Background
We expect strong hands-on engineering ability across several of these areas:
- Python
- Node.js / JavaScript
- REST APIs and webhooks
- SQL and database design
- MongoDB
- Docker/Linux
- Git
- n8n or similar orchestration
- Local/open-source LLM deployment
- Ollama, llama.cpp, vLLM, Hugging Face or similar
- OpenAI, Claude, Gemini or other model APIs
- RAG, embeddings and reranking
- Tool/function calling
- Structured outputs
- Agent state and memory
- Redis/Postgres/RDS or similar
- Monitoring and logging
You don't need experience with every technology above.
You do need enough engineering depth to learn unfamiliar systems and APIs independently.
n8n experience is useful. n8n-only experience is not enough.
This Role Is NOT For You If
Your AI experience is primarily:
- ChatGPT prompting
- No-code automation
- Zapier/Make workflows
- Basic n8n templates
- Tutorial chatbots
- AI wrappers
- Copy-paste RAG projects
- Front-end applications connected to an LLM API
We care far more about what you have actually built and how you think than how many AI tools appear on your résumé.
How to Apply
Along with your application, send us a 3–5 minute video/Loom walkthrough of the most sophisticated AI automation or agentic system you personally built.
Don't just show us the polished UI.
Show us:
- The problem you solved
- Architecture/workflow
- Actual code or orchestration
- Models used and why
- APIs and systems integrated
- State/memory approach
- How failures were handled
- What you personally built
- What didn't work initially
- What you would redesign today
If you've deployed local/open-source models, show us.
If you've reduced unnecessary LLM calls or token usage, explain how.
A chatbot, tutorial project, or downloaded n8n template by itself isn't sufficient.
Why BricksFolios
BricksFolios is building a technology-driven real estate wealth and property operations platform.
This role gives you something many AI engineering positions don't:
Real workflows to automate.
Real users.
Real operational problems.
And the ability to take systems from idea to production.
You won't be waiting for perfectly written engineering tickets.
You'll be expected to understand the business problem and figure out how technology can eliminate the work.
See a repetitive process.
Understand it.
Architect it.
Build it.
Deploy it.
Improve it.
That's the job.
Pay: Rs150,000.00 - Rs450,000.00 per month
Application Question(s):
Show us something real you built.
Share a link to a 3–5 minute video walkthrough of the most sophisticated AI automation or agentic system you personally built.
Show the actual workflow/code, integrations, what the AI did, and what you personally built.
A chatbot, tutorial project, or n8n template alone doesn't count.
- 3.
Architect across our systems without wasting tokens.
We use Follow Up Boss for CRM, ShowMojo for listing/tour management, Rentvine for property management, Twilio for communications, n8n for orchestration, plus our own Node.js/MongoDB/RDS platform.
How would you build a 24/7 agentic layer across these systems while avoiding repeatedly sending entire customer histories to an LLM?
Explain your approach to APIs/webhooks, state, memory, deterministic logic, local models, frontier models, retries, and human escalation.
- 2.
Design our 24/7 property-management agent.
A tenant emails at 11:30 PM: “There is water coming through the ceiling. I tried calling and nobody answered.” How would you design an end-to-end system to handle this without someone monitoring the inbox?
Explain how it determines urgency, retrieves tenant/property context, communicates with the tenant, contacts vendors, follows through to resolution, updates the property-management system, and decides when to involve a human.
Be specific about what you would handle with deterministic logic, local LLMs, frontier LLMs, n8n, and structured state.
- 4.
Design a 24/7 agentic lead-generation system.
We want a system that continuously finds prospects matching our ICP, enriches them, researches each prospect, identifies intent signals, scores them, generates personalized outreach, runs email/SMS sequences, understands replies, follows up intelligently, books meetings, and keeps Follow Up Boss updated.
How would you architect this end to end?
Be specific about what you would build with APIs/code vs. n8n, what should be deterministic, where local LLMs would be sufficient, when you would use a frontier model, how you would maintain prospect state without repeatedly sending full histories to an LLM, and how the system would learn from campaign results to improve targeting and conversion.
- Your night-shift track record. Where you worked it before, your hours, how long you sustained it, and whether 10 AM to 7 PM Pacific, six days a week, works for you long-term.
- Current monthly compensation in PKR
- Expected monthly compensation in PKR
- Earliest start date
Education:
- Bachelor's (Preferred)
Experience:
- AI Engineering: 2 years (Required)
Work Location: Remote
Find AI, ML, Data Science Jobs By Location
Find Jobs By Position