Software Developer Β· Applied AI Engineer
Software Developer at Energy One building enterprise energy-trading systems, and an MSc Applied AI candidate at the University of Warwick (WMG). I work across LLM systems, RAG pipelines, backend engineering and cloud deployment β turning operational problems into reliable, well-evaluated technical solutions.
π usmanmateen.com Β· πΌ LinkedIn Β· βοΈ jobs@usmanmateen.com
- Applied AI β LLM systems, RAG pipelines, prompt-injection guardrails, model evaluation and explainability
- Backend engineering β Python, C#/.NET, FastAPI, REST APIs, SQL, systems integration
- Cloud & delivery β GCP Cloud Run, AWS, Docker, CI/CD, Azure DevOps
- Responsible AI β governance, bias evaluation, safe deployment, human oversight
SentryML β LLM Guardrail & AI Security System
An AI firewall that detects LLM prompt-injection attacks. Semantic detection plus an optional fine-tuned DistilBERT classifier, ~95% detection accuracy on benchmarking across prompt injection, jailbreak, role hijacking and data-exfiltration. FastAPI inference on GCP Cloud Run with Firestore, SHAP explainability, a React monitoring dashboard, and published Python & JavaScript SDKs.
Python FastAPI GCP Cloud Run Firestore DistilBERT SHAP React
Enterprise RAG Evaluation β MSc Dissertation
An evaluation framework comparing sparse (BM25), dense and hybrid (RRF) retrieval with cross-encoder reranking on an enterprise financial-document benchmark. Measures groundedness, source attribution and hallucination risk, with failure analysis across retrieval and generation stages to recommend reliable deployment configurations.
Python RAG BM25 Dense Retrieval Reranking LLM Evaluation
Migrayt β Enterprise SaaS Migration Platform
A multi-tenant SaaS platform automating enterprise workflow migration between Azure DevOps and Jira. Connector-based architecture handling cross-platform data movement, transformation and sanitisation, with tenant isolation and access controls.
AWS Multi-tenant SaaS Connectors CI/CD
Paggio β Webhook Debugging & Replay Tool
A developer tool for diagnosing webhook failures β inspecting requests, reviewing payloads, isolating faults and replaying deliveries. Owned end to end from API design through testing and documentation.
API Design Webhooks Developer Tooling
Languages β Python Β· TypeScript Β· JavaScript Β· SQL Β· C# AI & ML β LLM APIs Β· RAG Β· embeddings Β· DistilBERT fine-tuning Β· model evaluation Β· SHAP Β· responsible AI Backend & data β FastAPI Β· REST APIs Β· SQL Server Β· PostgreSQL Β· data pipelines Β· webhooks Cloud & DevOps β GCP (Cloud Run, Firestore) Β· AWS Β· Docker Β· Azure DevOps Β· CI/CD Β· Git Frameworks β React Β· Node.js Β· Next.js Β· .NET Β· ASP.NET
MSc Applied Artificial Intelligence β University of Warwick (WMG) Β· 2025β2026 Responsible AI, AI governance, risk assessment, explainability and LLM evaluation.
BSc Computer Science with Artificial Intelligence β Coventry University Β· First-Class Honours Programming & Algorithms (91%), OOP (87%), Software Engineering (82%).
π« Open to conversations about applied AI and backend engineering β reach me at jobs@usmanmateen.com



