AI Agent & Automation Engineer
I build AI systems that increase revenue and reduce manual work, including lead capture and qualification, customer support agents, workflow automation, knowledge systems, and reporting dashboards.
Production-ready, integrated into your stack, delivered in days.
Over 15 production-grade AI systems built and deployed. Every project ships with evaluation pipelines, error handling, Docker deployment, and full source code handover.
Build stack:
Python, FastAPI, LangChain, LangGraph, n8n, Zapier, Make, Power BI • Built for production
CORE SERVICES
Production AI Systems Built Around Business Outcomes
Five production services covering the full AI automation stack, from revenue generation and customer support to internal operations, knowledge access, and reporting. Each system integrates into your existing tools and ships with working code, tests, and documentation.
REVENUE SYSTEMS
AI Lead Capture & Sales Automation
Capture, qualify, score, and route inbound leads automatically so your team responds faster and focuses on the best opportunities.
Last build: 80% faster lead response time. Configurable scoring rubric. Live on real data in 7 days.
Key Details:
Every lead scored within minutes
High-fit leads routed to reps via Slack + CRM update automatically
Follow-up triggers if no action is taken within your configured window
Enrichment, deduplication, and full audit trail
CUSTOMER OPERATIONS
AI Customer Support & Service Agents
Deploy AI support systems that answer from your company’s own documents, classify and route tickets, and escalate complex cases safely.
Last build: 65% faster first response on routine tickets. Live on real ticket data in 10 days.
Key Details:
Classifies every ticket by intent, urgency, and sentiment automatically
Answers routine questions from your own documents with source citations
Complex or urgent cases escalated with full context, never left unhandled
SLA tracking, Slack alerts, and weekly compliance report
OPERATIONS AUTOMATION
AI Workflow & Process Automation
Automate repetitive workflows involving emails, forms, documents, approvals, routing, and system updates, with structured extraction, validation, and human fallback paths.
Last build: 60% less manual processing. Near-zero routing errors. Deployed in 5 days.
Key Details:
Extracts structured data from emails, forms, PDFs, and file uploads
Validates, deduplicates, and routes to CRM, Slack, or spreadsheet
Confidence scoring flags uncertain items for human review
Runs via custom Python backend or n8n / Zapier / Make.com + custom AI
KNOWLEDGE SYSTEMS
AI Knowledge Base & Internal Assistant
Give your team instant access to company knowledge through an internal assistant that answers from your own documents with citations, guardrails, and measurable retrieval quality.
Last build: >85% retrieval accuracy. 50-question evaluation pipeline. Admin panel included.
Key Details:
Ingests your documents (PDF, Word, Markdown, Notion), instantly searchable
Every answer cited with document, page, and section, fully verifiable
Refuses to answer if evidence isn't there, no hallucination
Accuracy report on real Q&A from your domain delivered before handover
DATA & REPORTING
Automated Reporting & Intelligence Dashboards
Replace manual reporting with structured data pipelines and live dashboards that keep KPIs current, consistent, and decision-ready.
Last build: 8 hours/week of manual Excel reporting → 15-minute automated refresh.
Key Details:
Python pipeline connects your sources, cleans data, and feeds Power BI automatically
KPI logic in editable config, no code changes needed to update what you track
Threshold alerts via Slack or email when a KPI breaches a set limit
Refresh runbook included - your team can maintain it without a developer
DATA & REPORTING
Automated Reporting & Intelligence Dashboards
Replace manual reporting with structured data pipelines and live dashboards that keep KPIs current, consistent, and decision-ready.
Key Details:
Multi-source data integration
KPI logic and reporting automation
Dashboard refresh, alerts, and quality checks
CORE SERVICES
Production AI Systems Built Around Business Outcomes
Five production services covering the full AI automation stack, from revenue generation and customer support to internal operations, knowledge access, and reporting. Each system integrates into your existing tools and ships with working code, tests, and documentation.
REVENUE SYSTEMS
AI Lead Capture & Sales Automation
Capture, qualify, score, and route inbound leads automatically so your team responds faster and focuses on the best opportunities.
Last build: 80% faster lead response time. Configurable scoring rubric. Live on real data in 7 days.
Key Details:
Every lead scored within minutes
High-fit leads routed to reps via Slack + CRM update automatically
Follow-up triggers if no action is taken within your configured window
Enrichment, deduplication, and full audit trail
CUSTOMER OPERATIONS
AI Customer Support & Service Agents
Deploy AI support systems that answer from your company’s own documents, classify and route tickets, and escalate complex cases safely.
Last build: 65% faster first response on routine tickets. Live on real ticket data in 10 days.
Key Details:
Classifies every ticket by intent, urgency, and sentiment automatically
Answers routine questions from your own documents with source citations
Complex or urgent cases escalated with full context, never left unhandled
SLA tracking, Slack alerts, and weekly compliance report
OPERATIONS AUTOMATION
AI Workflow & Process Automation
Automate repetitive workflows involving emails, forms, documents, approvals, routing, and system updates, with structured extraction, validation, and human fallback paths.
Last build: 60% less manual processing. Near-zero routing errors. Deployed in 5 days.
Key Details:
Extracts structured data from emails, forms, PDFs, and file uploads
Validates, deduplicates, and routes to CRM, Slack, or spreadsheet
Confidence scoring flags uncertain items for human review
Runs via custom Python backend or n8n / Zapier / Make.com + custom AI
KNOWLEDGE SYSTEMS
AI Knowledge Base & Internal Assistant
Give your team instant access to company knowledge through an internal assistant that answers from your own documents with citations, guardrails, and measurable retrieval quality.
Last build: >85% retrieval accuracy. 50-question evaluation pipeline. Admin panel included.
Key Details:
Ingests your documents (PDF, Word, Markdown, Notion), instantly searchable
Every answer cited with document, page, and section, fully verifiable
Refuses to answer if evidence isn't there, no hallucination
Accuracy report on real Q&A from your domain delivered before handover
DATA & REPORTING
Automated Reporting & Intelligence Dashboards
Replace manual reporting with structured data pipelines and live dashboards that keep KPIs current, consistent, and decision-ready.
Last build: 8 hours/week of manual Excel reporting → 15-minute automated refresh.
Key Details:
Python pipeline connects your sources, cleans data, and feeds Power BI automatically
KPI logic in editable config, no code changes needed to update what you track
Threshold alerts via Slack or email when a KPI breaches a set limit
Refresh runbook included - your team can maintain it without a developer
DATA & REPORTING
Automated Reporting & Intelligence Dashboards
Replace manual reporting with structured data pipelines and live dashboards that keep KPIs current, consistent, and decision-ready.
Key Details:
Multi-source data integration
KPI logic and reporting automation
Dashboard refresh, alerts, and quality checks
FEATURED BUILDS
Production Systems & Representative Client Builds
A selection of production-focused systems built to reduce operational friction, improve speed and accuracy, and integrate cleanly into business workflows.
OPERATIONS AUTOMATION
Guardrailed Workflow Automation with Human Oversight
Client challenge
Operations teams were processing high volumes of inbound requests manually through email and spreadsheets, with inconsistent data quality, no audit trail, and too much dependency on human review.
What I built
A workflow automation system with structured extraction, validation rules, confidence scoring, exception routing, and human review for low-certainty cases.
System scope:
Ingestion from inbound operational channels
Structured extraction with schema validation
Confidence scoring and review routing
Audit logging and downstream notifications
Operational impact:
Reduced manual handling time
Improved output consistency
Clear escalation path for edge cases
Guardrails to prevent silent failures
Tech stack
Python • FastAPI • LLM API • Pydantic • PostgreSQL • Slack
OPERATIONS AUTOMATION
Guardrailed Workflow Automation with Human Oversight
Client challenge
Operations teams were processing high volumes of inbound requests manually through email and spreadsheets, with inconsistent data quality, no audit trail, and too much dependency on human review.
What I built
A workflow automation system with structured extraction, validation rules, confidence scoring, exception routing, and human review for low-certainty cases.
System scope:
Ingestion from inbound operational channels
Structured extraction with schema validation
Confidence scoring and review routing
Audit logging and downstream notifications
Operational impact:
Reduced manual handling time
Improved output consistency
Clear escalation path for edge cases
Guardrails to prevent silent failures
Tech stack
Python • FastAPI • LLM API • Pydantic • PostgreSQL • Slack
REVENUE AUTOMATION
AI Lead Processing & CRM Routing System
Client challenge
Inbound leads were arriving from forms and campaigns, but qualification and routing were inconsistent. Good opportunities were sitting too long, and sales teams were spending time on low-fit leads.
What I built
A lead processing system that captures inbound submissions, enriches and scores them, applies routing logic, and pushes high-fit opportunities into the right CRM and internal workflows.
System scope:
Lead ingestion from forms and campaigns
AI scoring against configurable criteria
CRM routing and internal alerts
Follow-up automation and deduplication
Operational impact:
Faster lead response times
More consistent qualification
Better handoff from marketing to sales
Lower manual triage overhead
Tech stack
Python • FastAPI • Zapier / n8n • CRM • Slack • Google Sheets
REVENUE AUTOMATION
AI Lead Processing & CRM Routing System
Client challenge
Inbound leads were arriving from forms and campaigns, but qualification and routing were inconsistent. Good opportunities were sitting too long, and sales teams were spending time on low-fit leads.
What I built
A lead processing system that captures inbound submissions, enriches and scores them, applies routing logic, and pushes high-fit opportunities into the right CRM and internal workflows.
System scope:
Lead ingestion from forms and campaigns
AI scoring against configurable criteria
CRM routing and internal alerts
Follow-up automation and deduplication
Operational impact:
Faster lead response times
More consistent qualification
Better handoff from marketing to sales
Lower manual triage overhead
Tech stack
Python • FastAPI • Zapier / n8n • CRM • Slack • Google Sheets
KNOWLEDGE SYSTEMS
Internal AI Knowledge Assistant
Client challenge
Teams were wasting time searching across internal documents, SOPs, and reference material, or relying on specific people to answer questions manually.
What I built
An internal knowledge assistant that retrieves from company documents, generates grounded answers with citations, and refuses unsupported responses when evidence is insufficient.
System scope:
Multi-format document ingestion
Retrieval and answer generation pipeline
Citation-backed responses
Evaluation workflow and admin controls
Operational impact:
Faster access to internal knowledge
More consistent answers across teams
Reduced dependency on tribal knowledge
Safer AI behaviour through grounding and guardrails
Tech stack
Python • FastAPI • LangChain • Vector Database • OpenAI / Claude API • Docker
KNOWLEDGE SYSTEMS
Internal AI Knowledge Assistant
Client challenge
Teams were wasting time searching across internal documents, SOPs, and reference material, or relying on specific people to answer questions manually.
What I built
An internal knowledge assistant that retrieves from company documents, generates grounded answers with citations, and refuses unsupported responses when evidence is insufficient.
System scope:
Multi-format document ingestion
Retrieval and answer generation pipeline
Citation-backed responses
Evaluation workflow and admin controls
Operational impact:
Faster access to internal knowledge
More consistent answers across teams
Reduced dependency on tribal knowledge
Safer AI behaviour through grounding and guardrails
Tech stack
Python • FastAPI • LangChain • Vector Database • OpenAI / Claude API • Docker
BESPOKE SERVICES
Custom Engineering Solutions
Need something beyond the standard sprints? I offer bespoke development for unique technical challenges
BESPOKE SERVICES
Custom Engineering Solutions
Need something beyond the standard sprints? I offer bespoke development for unique technical challenges
WHY CLIENTS HIRE ME
Built for Production, Not Just Demo Environments
The difference between a working demo and a production-ready system is everything around the model: validation, fallback logic, observability, error handling, evaluation, and integration into the way a team actually works. That is the layer I focus on.
Production-first engineering
Systems are built to handle real inputs, edge cases, retries, failures, and operational handoffs.
Structured outputs and validation
Where AI is used for extraction, scoring, or classification, outputs are constrained, validated, and checked before they trigger downstream actions.
Human fallback paths
Low-confidence or ambiguous cases can be routed to a review queue rather than forcing unreliable automation.
Evaluation and measurable quality
Systems include testing and evaluation so performance is measured, monitored, and improved over time.
Clean integrations
Solutions are designed to fit into the tools teams already use, including CRMs, Slack, spreadsheets, dashboards, and internal platforms.
Clear handover
Projects are delivered with documentation, deployment guidance, and a structure your team can maintain after handoff.
WHY CLIENTS HIRE ME
Built for Production, Not Just Demo Environments
The difference between a working demo and a production-ready system is everything around the model: validation, fallback logic, observability, error handling, evaluation, and integration into the way a team actually works. That is the layer I focus on.
Production-first engineering
Systems are built to handle real inputs, edge cases, retries, failures, and operational handoffs.
Structured outputs and validation
Where AI is used for extraction, scoring, or classification, outputs are constrained, validated, and checked before they trigger downstream actions.
Human fallback paths
Low-confidence or ambiguous cases can be routed to a review queue rather than forcing unreliable automation.
Evaluation and measurable quality
Systems include testing and evaluation so performance is measured, monitored, and improved over time.
Clean integrations
Solutions are designed to fit into the tools teams already use, including CRMs, Slack, spreadsheets, dashboards, and internal platforms.
Clear handover
Projects are delivered with documentation, deployment guidance, and a structure your team can maintain after handoff.
ENGAGEMENT PROCESS
How Projects Typically Work
Most projects are delivered as fixed-scope builds with a clear problem definition, scoped implementation, and clean handover. Ongoing contract support is available where embedded collaboration is needed.
Step 1
Define the problem
We identify the business function, workflow, or system that needs to be improved, along with the desired outcome, constraints, and tools involved.
Step 2
Scope the right solution
I map the implementation approach based on your workflow, risk tolerance, team setup, and existing tools, whether that means a custom backend, orchestration layer, internal assistant, or reporting pipeline.
Step 3
Build and validate
The system is built with the required integrations, business rules, validation logic, and review paths, then tested against representative scenarios.
Step 4
Deploy and hand over
Once the system is working as intended, it is deployed, documented, and handed over with the information your team needs to operate it confidently.
Available for fixed-scope projects with defined deliverables, or ongoing contract engagements where embedded collaboration is needed.
Step 1
Discovery Call (15–30 minutes)
We discuss your pain point, review examples, and confirm scope. If we're aligned, you receive a fixed-price proposal with clear deliverables and timeline.
Step 3
Build & Iteration (Days 2–8)
Rapid development with regular check-ins. You review working prototypes and provide feedback. Changes within scope are included; anything beyond triggers a change control discussion.
Step 2
Project Kickoff (Day 1)
Access setup, requirement confirmation, and stakeholder alignment. We document definitions, workflows, and acceptance criteria before any build work begins.
Step 4
Handover & Training
Complete documentation, Loom walkthrough, and live training session. You receive all code, credentials, runbooks, and ongoing operational guidance.
ENGAGEMENT PROCESS
How Projects Typically Work
Most projects are delivered as fixed-scope builds with a clear problem definition, scoped implementation, and clean handover. Ongoing contract support is available where embedded collaboration is needed.
Step 1
Define the problem
We identify the business function, workflow, or system that needs to be improved, along with the desired outcome, constraints, and tools involved.
Step 2
Scope the right solution
I map the implementation approach based on your workflow, risk tolerance, team setup, and existing tools, whether that means a custom backend, orchestration layer, internal assistant, or reporting pipeline.
Step 3
Build and validate
The system is built with the required integrations, business rules, validation logic, and review paths, then tested against representative scenarios.
Step 4
Deploy and hand over
Once the system is working as intended, it is deployed, documented, and handed over with the information your team needs to operate it confidently.
Available for fixed-scope projects with defined deliverables, or ongoing contract engagements where embedded collaboration is needed.
Step 1
Discovery Call (15–30 minutes)
We discuss your pain point, review examples, and confirm scope. If we're aligned, you receive a fixed-price proposal with clear deliverables and timeline.
Step 3
Build & Iteration (Days 2–8)
Rapid development with regular check-ins. You review working prototypes and provide feedback. Changes within scope are included; anything beyond triggers a change control discussion.
Step 2
Project Kickoff (Day 1)
Access setup, requirement confirmation, and stakeholder alignment. We document definitions, workflows, and acceptance criteria before any build work begins.
Step 4
Handover & Training
Complete documentation, Loom walkthrough, and live training session. You receive all code, credentials, runbooks, and ongoing operational guidance.
BACKGROUND & STACK
BACKGROUND & STACK
Technical Depth with Practical Delivery Focus
Technical Depth with Practical Delivery Focus
I work across the application, orchestration, and system layers of AI automation, from workflow tooling and integrations through to custom Python services, retrieval systems, and multi-step agent logic.
MSc Computer Science (AI/ML specialisation)
Chemical Engineering (IChemE-accredited)
Previous platform security engineering experience
Tech stack
Python, FastAPI, LangChain, LangGraph, n8n, Zapier, Make.com, OpenAI / Claude API, ChromaDB, PostgreSQL, Power BI, Docker.
Also well suited to process-driven environments where traceability, accuracy, and domain context matter.
Contact Me
Let's connect
