05·AI / ML
Barnes AI
Domain LLM for Dubai luxury real estate. Fine-tuned (QLoRA) on ~9,600 instruction pairs sourced from Dubai Land Department, UK Land Registry, France DVF, Singapore URA, and the World Bank. Deployed via Ollama with zero per-query cost.
What it took
The skills behind this project.
Every project below leans on a primary discipline and a handful of secondary ones. Tap any chip to see how that skill plays out across the wider portfolio.
Skills demonstrated
- AI & Machine LearningPrimary
QLoRA fine-tune of a domain LLM plus a multi-agent reason-propose-approve-execute loop.
- Backend EngineeringSecondary
FastAPI + Ollama deployment with audit-logged human-in-the-loop above the confidence threshold.
- Data PipelinesSecondary
9,600 instruction pairs sourced from DLD, UK Land Registry, France DVF, Singapore URA, and World Bank.
- ResearchSupporting
Training-run analysis and dataset assembly that targeted brokerage-grade jurisprudence.
Context
Why it exists.
Atlas surfaces the data; Barnes AI is what reasons over it. A multi-agent research system fine-tuned on real-estate jurisprudence, deployed locally via Ollama with zero per-query cost, capable of producing brokerage-grade analyses across ten-plus sources on demand.
The core decision was to fine-tune rather than wrap a frontier model with prompts. Brokerage-grade output needs a model that understands DLD transaction conventions, France DVF reporting cadence, Singapore URA mapping — none of that lives in a frontier model's pre-training in a useful way. Nine thousand six hundred instruction pairs, hand-curated from five jurisdictions, taught the base model what BARNES Dubai actually means when it says `recent`, `comparable`, or `flag for review`.
StackPyTorch · QLoRA · Ollama · Python · FastAPI · Multi-agent
Process
The decisions that shaped it.
- 01
QLoRA on a domain-specific corpus, not RLHF
QLoRA over a base instruct model was the right tier — full fine-tune was overkill, prompt engineering was undertrained. The instruction pairs were sourced from Dubai Land Department, UK Land Registry, France DVF, Singapore URA, and the World Bank; each pair was reviewed by an analyst so the model learns the team's judgment, not just the public registry's wording.
AI & Machine Learning - 02
Local deployment via Ollama, not a hosted endpoint
Per-query cost on a hosted frontier API would have foreclosed the on-demand research workflow before it shipped. Local Ollama deployment on a single workstation drives the cost to electricity, runs at conversational latency, and keeps all transactional data inside the org's network — non-negotiable for a brokerage handling proprietary listings.
Backend Engineering - 03
Victor: a seven-layer event loop, not chat completion
The multi-agent wrapper (Victor) treats each research run as Event → Evaluate → LLM Reason → Propose → Approve → Execute → Log. Above the confidence threshold the agent commits; below it, the proposal escalates to a human. The audit log feeds the next training pass — accepted and rejected actions are both signal.
AI & Machine Learning - 04
FastAPI in front so other tools can call the agents
Wrapped Victor behind a FastAPI service so Atlas, the Innovation Portal, and ad-hoc Python scripts all hit the same agents through one API. Means the multi-agent system is a shared brain across the BARNES Dubai stack, not a one-off chat surface.
Backend Engineering
Outcome
What shipped.
Multi-agent research system ingests ten+ sources (transaction registries, listing portals, social, news, CRM) and produces ranked opportunities on demand.
In production with the BARNES Dubai research team. Ten-plus ingest sources — transaction registries, listing portals, social, news, internal CRM — flow through Victor and produce ranked opportunities on demand. The audit trail closes the loop: the next training pass learns what the team chose to act on.