Agency-governed legal intelligence platform for law enforcement operations.
Built LEO as a multi-surface AI platform spanning public intake, officer workspace, agency administration, and platform operations, with agentic RAG, cited legal answers, credit governance, answer-quality review, support workflows, and knowledge-base operations.
Role
Founder, product architect, full-stack AI platform builder
Built four connected product surfaces: public website, officer workspace, agency portal, and platform admin control plane.
Implemented an agentic RAG workflow with intent detection, conversation-aware retrieval, legal-topic expansion, citations, and controlled Deep Research.
Created credit governance with agency wallets, officer allocations, reservations, ledger tracking, usage analytics, and billing safeguards.
Problem
What was broken or inefficient?
Law enforcement agencies need more than a chatbot: they need source-grounded AI that officers can use safely while administrators control access, spend, documents, support, and answer quality.
Solution
What Mohammed built
LEO AI packages legal intelligence into a governed, multi-role platform where officers get cited answers and report-ready workflows while agencies and platform admins manage credits, permissions, support, knowledge operations, and auditability.
Execution
What Mohammed built to move the product forward.
LEO AI packages legal intelligence into a governed, multi-role platform where officers get cited answers and report-ready workflows while agencies and platform admins manage credits, permissions, support, knowledge operations, and auditability.
Multi-surface platform architecture
Officer AI workspace with cited answers and saved responses
Agency admin for officers, credits, documents, support, and settings
Platform admin control plane for onboarding, billing, knowledge operations, and answer quality
Agentic RAG with controlled Deep Research
Credit reservation, usage ledger, support, and audit workflows
The live LEO AI site presents the platform as an agency-grade product, connecting public intake with officer access, cited answers, and administrative control.
These visuals come from the official public-facing project surfaces and help ground the case study in the actual product experience, not just portfolio copy.
Execution Surface
Core modules and product decisions
The strongest work sits at the intersection of product judgment, technical architecture, and operating constraints.
Built four connected product surfaces: public website, officer workspace, agency portal, and platform admin control plane.
02
Implemented an agentic RAG workflow with intent detection, conversation-aware retrieval, legal-topic expansion, citations, and controlled Deep Research.
03
Created credit governance with agency wallets, officer allocations, reservations, ledger tracking, usage analytics, and billing safeguards.
04
Added operational oversight through support workflows, answer-quality review, audit logs, knowledge-base administration, and agency settings.
01
Multi-surface platform architecture
02
Officer AI workspace with cited answers and saved responses
03
Agency admin for officers, credits, documents, support, and settings
04
Platform admin control plane for onboarding, billing, knowledge operations, and answer quality
05
Agentic RAG with controlled Deep Research
06
Credit reservation, usage ledger, support, and audit workflows
Outcomes
Repositioned LEO as a managed enterprise AI platform rather than a prototype chatbot.
Connected officer workflow, agency administration, platform operations, and monetization into one product system.
Made legal AI usage more governable through citations, credits, role-based controls, quality review, support, and auditability.
Lessons
Regulated AI products need an operating layer around the model: access, billing, support, quality, and knowledge maintenance.
Cost governance and answer-quality workflows are product trust features, not back-office extras.
Recruiter Takeaway
Mohammed can turn regulated AI into a complete SaaS operating platform: officer UX, admin control, agentic retrieval, billing, governance, and operational oversight.
Architecture
How The System Holds Together
These are public-safe architecture layers: enough to show leadership judgment without exposing sensitive implementation detail.
01
Agentic RAG Core
Conversation-aware retrieval, legal-topic expansion, retrieval planning, approved-source grounding, and citation storage shape the answer workflow.
02
Governance Plane
Officer, agency, and platform-admin surfaces coordinate permissions, agency settings, support escalation, answer review, and audit trails.
03
Usage Economy
Agency wallets, officer allocations, reservations, ledger entries, usage analytics, and billing controls make AI spend governable.
