Esther Usman

Esther Usman

I make the Impossible possible

I build in n8n, Claude Code, and custom code, and I add the error handling, logging, and documentation that keep a system running after handoff.

Get in touch Open to remote, full-time roles
FINTECHTrustBridge
CRMOmniStore
RESEARCHProtocolPulse
CONTENT OPSLinkedIn Content Manager
WEB3SAVA
SUPPORTEmber & Co.
RAGNorthstar RAG Assistant
2024
Started building AI automation

What I do

Workflow automation

Lead routing, data entry, follow ups, and internal handoffs that run without anyone managing them. In OmniStore, every new lead is scored, answered in under a minute, and logged to the CRM with no manual entry.

Example: OmniStore CRM and Lead Automation

AI agents

Agents that read your data, make a decision, and act inside the tools you already use. TrustBridge runs credit scoring and onboarding for SME applicants and cut manual processing by an estimated 70%.

Example: TrustBridge

Custom development

When a no code tool cannot express the logic, I write it in Python, JavaScript, or TypeScript: scoring models, backend integrations, and assistants that answer from your own documents.

Example: Northstar RAG Assistant

How I work

Scope & discovery

Understand the problem, the data involved, and what success looks like before choosing a tool.

Smallest working version

Build the smallest version that proves the workflow actually works, then iterate from there.

Reliability layer

Add error handling, logging, and documentation so the system holds up after handoff.

Handoff

Deliver with clear instructions so the system keeps running without me in the loop.

Selected work

FINTECH

TrustBridge

Multi-agent system, built for the Band of Agents Hackathon on lablab.ai

A multi-agent system for alternative credit scoring, built to support onboarding and internal operations for SME applicants without traditional credit histories.

In
  • Applicant and SME data
Process
  • 6+ orchestrated agent workflows
  • 4+ APIs for data and decisioning
  • Alternative scoring logic
Out
  • About 70% less manual processing
  • Consistent onboarding decisions
Multi-agent orchestrationCredit scoring logicFintech APIs
CRM

OmniStore CRM & Lead Automation

End to end intake to CRM pipeline built in n8n

Reads customer interest, scores buying intent, and follows up automatically. Tested against 50+ simulated leads with zero manual data entry.

In
  • Customer interest form
Process
  • AI scores intent, 0 to 100
  • Classifies Hot, Warm, or Cold
  • Logs 100% of leads to CRM
Out
  • AI response in under 1 minute
  • Follow up triggers for hot leads
n8nAI ScoringGoogle Sheets CRMAutomated Email
RESEARCH

ProtocolPulse

Autonomous research agent built on Dify.ai

Turns raw protocol data into standardized briefs, using 8+ workflow modules and iteration loops for consistency across runs.

In
  • Protocol or topic to research
Process
  • Live web research
  • 8+ automated workflow modules
  • Iteration loops for accuracy
Out
  • Standardized research briefs
  • 10+ protocols covered in testing
Dify.aiGPT-4oLive Web Research
CONTENT OPS

LinkedIn Content Manager

Content operations agent built end to end in Claude Code

Turns a single topic into a ready to publish LinkedIn post, then manages it through calendar, approval, and publish tracking.

In
  • A topic or idea prompt
Process
  • AI drafts the post
  • Logs to content calendar
  • Tracks Draft, Ready, Published
Out
  • Manages 30+ posts end to end
  • Publication tracking
Claude CodeMCPFilesystem Integration
WEB3 / SAVINGS

SAVA

AI native Ajo/Esusu savings agent on Solana, built in Claude Code

Handles group creation, contributions, and payout rotation on chain via Solana Pay, with AI driven member reliability scoring.

In
  • Group creation and member contributions
Process
  • Solana Pay contribution handling
  • Payout rotation logic
  • AI member reliability scoring
Out
  • Automated payout to group members
  • On chain, end to end tested
Claude CodeSolana web3.jsSolana PayExpress / TypeScript
SUPPORT

Ember & Co.

Conversational support agent built in Voiceflow

Automates common booking and FAQ flows before handing off to a human. Validated across 20+ simulated customer conversations.

In
  • Customer message
Process
  • Knowledge base lookup
  • Booking and FAQ conversation flows
Out
  • Resolved in flow, or escalated
  • Validated on 20+ conversations
VoiceflowKnowledge Base DesignConversation Flows
RAG

Northstar RAG Assistant

Document question answering assistant built with retrieval augmented generation

Answers questions about an employee handbook using only what the document says, by retrieving the relevant passages first and then generating a reply from them.

In
  • PDF documents and user questions
Process
  • Overlapping chunking and embeddings
  • Semantic retrieval from ChromaDB
  • Context based answer generation
Out
  • Answers grounded in the source document
  • Chatbot interface, no paid API key needed
PythonChromaDBSentence TransformersHugging FaceGradio

Skills

Automation platforms
n8nMakeZapierGoHighLevel
AI & agents
Claude CodeClaude / OpenAI APIDify.aiVoiceflowModel Context Protocol (MCP)RAG pipelines
Custom development
JavaScript / TypeScriptPythonNode.js / ExpressSolana web3.jsChromaDB / vector databases
Data & integration
Google SheetsGoogle FormsGmail APIAirtableNotion

Education

B.Sc. (Ed), Human Kinetics
University of Ilorin
National Diploma, Science Laboratory Technology
Kwara State Polytechnic

Certifications

Master AI Automation Certification
Udemy, Build Real AI Agents from Scratch
May 2026
n8n Masterclass 2026
Udemy
October 2026
Level-up Entrepreneurship Skills Course
Unilever
Certified

Community

Social Media Manager (Volunteer)
LinkedIn Local Ilorin, Science & Technology
Ongoing

Let's work together.

I'm an AI Automation Engineer open to remote, full time roles. If you need someone who can automate a workflow, build an agent, or write custom code when a no code tool runs out of road, I'd like to hear about it.