UM.

AI Engineer · Birmingham, UK

Shipping RAG systems end to end.

I build and ship production AI. LLMs, RAG, and systems that run in production.

> open to mid-level ai / applied-ai roles

build · ship
Developer ID

Applied AI · 2025

UM.
Engineering passClearance · L3
Usman Mateen
Usman Mateen
AI Engineer
LLM · RAG · Production systems
ID ENG-2025-0847
ISS Mar 2025
EXP Dec 2026
Birmingham, UK
Active · On-site
NFC
USMANMATEEN.COM

SCAN

Issued by usmanmateen.com · WMG · Energy One

CV

PDF · tap to open

Usman Mateen

AI Engineer

Mid-level AI roles · Birmingham

Now playing
…
…
connecting
Birmingham, UK
--:--:--
GMT
Europe/London
Case study1 / 4OneSpotMicrosoft Teams app10 claims at once · exactly 1 bookingRead the case study →

OneSpot: 10 claims at once · exactly 1 booking

Education
MSc Applied AI
Warwick (WMG) · awaiting results
BSc CS with AI · First-Class
Coventry University
EN · UR · PT (spoken)
~/research
$ dissertation --status
› Hybrid RAG benchmark
dense · sparse · KG · visual
# FinanceBench
✓ submitted
awaiting grades
$
Stack
AI / ML
LLMsRAGPyTorchVector searchSHAPHugging Face
Languages
PythonC# / .NETTypeScriptSQLC++
Cloud & data
AWSGCPCloudflare WorkersTerraformPostgreSQLDocker
Ship
FastAPINext.jsNode.jsClaude APITeams + Entra IDAzure DevOps
UKBHX

Birmingham skyline

Greetings from

Birmingham, UK

52.48°N · open to remote & hybrid

Stamped
PKPTESNOUKGI
Find me online
LinkedIn GitHub
Selected work

Real projects, shipped. Not just slideware.

A security SDK, a commercial SaaS platform, a Microsoft Teams app, and products launched solo. Each solved a real problem for real users.

> SentryML_
NatWest × Google Cloud Hackathon · 2026

An AI firewall that detects LLM prompt-injection attacks in real time.

LLM-powered products are vulnerable to prompt injection, and there was no lightweight, explainable guardrail to catch it inline.

GitHubPyPInpmCase study
Result
97% accuracy · sub-millisecond latency

Designed and delivered in 48 hours, pitched to NatWest and Google Cloud judges. Deployed to GCP Cloud Run, published as an open-source Python SDK, with a React monitoring dashboard for real-time visibility.

Stack
PythonFastAPIDistilBERTSHAPGCP Cloud RunReact
migrayt logo
B2B SaaS · Founder · Ongoing

A commercial SaaS platform that automates project-data migration between Azure DevOps and Jira.

Migrating tickets and project data between enterprise tools is slow, manual, and error-prone.

Architected a secure, scalable, multi-tenant cloud platform (AWS, infrastructure-as-code) with an AI-driven data-processing layer under a strict enterprise data-privacy model, plus automated CI/CD, tenant isolation, and usage-based pricing.

LiveCase study
Stack
AWS (ECS, RDS, SQS, VPC)TerraformAnthropic APIPostgreSQLStripeAzure DevOps CI/CD
OneSpot
Microsoft Teams app · Workplace tool · 2026

A Microsoft Teams app where people release paid parking spaces they aren't using and colleagues claim them in one tap.

The office parking group chat couldn't stop two people taking the same space, kept no history, and couldn't say who paid for which bay.

One rule, one entitlement per person per day, enforced three times: in the domain, with a per-person row lock, and by PostgreSQL triggers. Identity comes from Microsoft Entra ID without Microsoft Graph, and notifications go out through a transactional outbox.

Case study
Result
10 simultaneous claims · exactly 1 booking

Built end to end: Adaptive Cards, a shared noticeboard, and a Parking Admin tab served by the same Node process. Piloted in a developer Microsoft 365 tenant, with 550+ automated tests.

Stack
TypeScriptNode.jsMicrosoft TeamsEntra IDPostgreSQLDockerAzure DevOps
PulseWatch.
Personal project · Wearables

A Cloudflare Worker that notices when my Fitbit isn't being worn and sends a Telegram reminder.

A seven-state machine reads the gap before the tracker's last sync, with 30-day personal baselines and a transactional outbox, for about £0 a month.

TypeScriptCloudflare WorkersD1QueuesTelegram Bot API
Case study
grabstays
Student Tenancy Platform · Founder

A verified platform for students to take over each other's tenancies safely.

Built and launched solo across two universities, owning market research, product, build, deployment and a zero-budget go-to-market across print, Reddit and student press.

Next.jsSupabaseRailway
LiveGitHubCase study
$ paggio_
Developer Tool · Founder

A tool for developers to debug and replay webhooks.

Built and publicly launched on Product Hunt, idea to live product.

ReactFastAPI
LiveProduct Hunt

PhoneHub

Point-of-Sale · Client Delivery

A point-of-sale and inventory system for a family-run phone-repair business.

Next.jsPostgreSQL
Details
Open source

Contributing to open-source projects.

Fixes and improvements I've contributed to open-source projects that other developers depend on. Updated automatically from GitHub.

GitHub achievements
Merged
3
In review
4
Projects
3
452 contributions in the last 20 weeks
Experience

Shipping in production, not just side projects.

Energy One

Software Developer

Energy One
Jul 2025 - Present
PythonC#/.NETSQL ServerREST APIsAzure DevOps
  • Scoped, built and evolved major parts of the Python tooling for a Zendesk-to-Jira migration of 10,000+ support tickets (~99.8% migration success). Designed and implemented per-ticket migration state and resume; implemented rolling six-ticket concurrency, a major part of a ~5.4× throughput improvement. Engineering case study →
  • Introduced AI- and automation-based tooling, including LLM/RAG solutions, to streamline internal workflows and reduce manual effort across teams.
  • Partner with business teams to translate commercial requirements into delivered solutions for enterprise energy-trading (ETRM) systems, owning work from requirements through design to deployment.
Energy One

Software Engineer Intern

Energy One
Sep 2024 - Jul 2025
PythonData analysisQA
  • Contributed to data processing and analysis for AI-based energy-market prediction models and supported feature development across the platform.
  • Built a strong understanding of how technical delivery connects to business outcomes through QA, support and load testing in a live trading environment.
Smarttel

Senior IT Support Engineer

Smarttel · Remote
Apr 2023 - Oct 2024
InfrastructureClient support
  • Resolved complex infrastructure issues for business clients with a 95% resolution rate, building client-facing communication and structured problem-solving under operational pressure.
Education

Computer Science first, then Applied AI.

University of Warwick

MSc Applied AI

University of Warwick · WMG
2025 - 2026
Dissertation submitted · awaiting results
RAGRetrieval evaluationAI governanceRisk assessment
  • Consultancy-style AI engagements for a simulated client across healthcare, finance and intelligent systems, producing AI adoption strategies, risk assessments and mitigation plans.
  • Led an ethical AI assessment of an NHS chest X-ray triage system, covering governance, bias and transparency.
  • Dissertation: benchmarking hybrid retrieval strategies (dense, sparse, knowledge-graph and visual-semantic) for enterprise RAG on the FinanceBench financial dataset.
Coventry University

BSc (Hons) Computer Science with AI

Coventry University
2022 - 2025
First-Class Honours
Machine learningForecastingFull-stack
  • Final-year project: an AI forecasting model built on live energy-market data.
  • Built a full-stack MRP system with role-based access, graded 82%.
  • Member of the AI Race Team.
About

Engineer first, AI second. both, always shipping.

I'm an AI engineer and software developer with a First-Class BSc in Computer Science with AI from Coventry, finishing an MSc in Applied AI at Warwick (WMG). I build applied AI (LLM and RAG systems, automation, integrations) and I ship my own products end to end.

Right now I'm at Energy One in the energy-trading sector, working across Python, C#/.NET and TypeScript. My dissertation benchmarks hybrid retrieval strategies (dense, sparse, knowledge-graph and visual-semantic) for enterprise RAG against the FinanceBench dataset.

I care about the unglamorous part: making AI actually run: reliable, explainable, and in production. And I can explain any of it to someone non-technical without the jargon.

Based in Birmingham and open to mid-level AI engineer, AI developer, and applied machine learning roles across the UK: remote, hybrid, or relocation considered for teams building LLM products and RAG systems.

~/currently
  • Buildingenterprise systems at Energy One (energy-trading / ETRM)
  • StudyingMSc Applied AI at Warwick (WMG)
  • Researchinghybrid retrieval for enterprise RAG vs. FinanceBench
Looking for

Mid-level AI / applied-AI / software-engineering roles.

Stack

The tools I reach for.

Languages & Frameworks

code
PythonC# / .NETASP.NETTypeScript / JavaScriptSQLC++FastAPIReactNext.jsNode.jsTailwind CSSZodDrizzle ORMVitest

AI & Machine Learning

ai/ml
LLM applicationsRetrieval-Augmented Generation (RAG)Hybrid retrieval (dense, sparse, knowledge graph)RAG evaluationPrompt engineeringNLPEmbeddings & vector search (FAISS, BM25)Fine-tuning (DistilBERT)Explainable AI (SHAP, Integrated Gradients)PyTorchTensorFlowscikit-learnHugging Face

AI Tooling & Agentic Dev

tooling
Claude APIAmazon BedrockModel Context Protocol (MCP)Custom MCP server (TypeScript)Claude CodeKiroSpec-driven developmentCursorWindsurfn8n

Cloud & DevOps

cloud
AWS (ECS, RDS, SQS, VPC, Secrets Manager)Azure DevOps (CI/CD)Google Cloud (Cloud Run)Cloudflare Workers (D1, Queues)TerraformDockerGitHub ActionsVercelGitFirebase & Firestore

Data & Integration

data
PostgreSQLSQL ServerMicrosoft Teams appsMicrosoft Entra ID (SSO)REST APIsWebhooksTransactional outboxTelegram Bot APIStripePostmanMigrations (Jira, Zendesk, ADO)

Consulting & Delivery

delivery
Problem scoping & diagnosticsRequirements gatheringStakeholder communicationAI adoption strategyRisk assessmentRapid prototypingAgile / Scrum
Contact

Open to mid-level AI roles. say hello.

Hiring an AI engineer or developer? The fastest way to reach me is email. I read everything and reply to anything that looks like a fit.

jobs@usmanmateen.com
LinkedIn GitHub