Backend engineer focused on AI evaluation systems, trading/market infrastructure, automation, and data-intensive backend platforms.

I work on systems that ingest messy real-world inputs, enforce constraints, preserve auditability, and produce outputs that can be tested, replayed, and inspected.

I currently freelance on evaluations for frontier AI systems, with a focus on agentic coding agents, professional workflow simulation, and model failure analysis.

This includes:

• Designing realistic evaluation scenarios for pre-release and frontier models.
• Creating supporting artifacts such as dummy CSVs, TXT files, configs, prompts, system instructions, and task briefs.
• Building workflow simulations around professional personas such as hospital operations managers, forensic lab coordinators, facilities planning specialists, compliance reviewers, and technical operators.
• Analysing failures in reasoning, tool use, execution, formatting, constraint handling, and final output quality.
• Designing Operations Research evaluations with OR-Tools, including solver scripts, golden responses, optimal-value JSONs, configuration verifiers, and failure analyses.

Previously, I worked for just under two years in Central London as an AI Automation Engineer and Trading Operations Manager at a simulated prop trading firm. The role covered trading platform operations, market data workflows, AI support automation, dispute operations, vendor escalation, and production-facing automation systems.

I am currently building across:

• Backend and platform engineering
• Agentic AI evaluation infrastructure
• Trading and market data systems
• Operations Research and constraint-solving workflows
• Cloud-backed automation systems
• Data pipelines and validation-heavy service layers

My engineering bias is toward systems that are:

• observable
• testable
• replayable
• auditable
• failure-aware
• explicit about the boundary between deterministic logic and probabilistic model behaviour

My current interests sit across four areas:

Backend systems, APIs, data services, service boundaries, CI/CD, testability, and maintainable architecture.

Optimization modelling, constraint encoding, solver-backed applications, optimal-output verification, and failure analysis in model-generated solver code.

System decomposition, requirements, interfaces, validation, operational risk, and reliability under real-world constraints.

Agent evaluation, tool-use reliability, prompt and artifact design, traceability, workflow automation, and deterministic/probabilistic hybrid systems.

Longer term, I am interested in formal methods, compiler engineering, and correctness in low-level or reliability-critical environments.

Experience across simulated trading infrastructure, operational tooling, and market data workflows, including:

• MetaTrader 5, cTrader, and MatchTrader administration
• Trading account configuration and lifecycle operations
• Market data ingestion and reconciliation
• OHLCV data pipelines
• Execution quality investigation
• Latency and platform incident analysis
• Operational risk controls
• Broker/vendor escalation workflows

Experience designing and operating applied AI systems for workflow automation, support operations, and model evaluation, including:

• RAG pipeline design
• Intent/entity modelling
• Guardrail design
• Knowledge-base restructuring
• AI support automation
• Deterministic + stochastic workflow design
• Agentic AI evaluation
• Prompt artifact construction
• Failure-mode analysis
• Traceable and auditable AI-assisted workflows

I am especially interested in using AI as a bounded system component: useful where it reduces operational load, but still measurable, constrained, and supported by fallback paths.

Python, SQL, HTML5, CSS, Bash, PowerShell

FastAPI, Flask, Django, REST APIs, WebSockets, JSON-RPC, GraphQL, RSS

PostgreSQL, SQLite, Microsoft SQL Server, Redis, Azure Storage, AWS S3, Pandas, Polars, PyArrow, Apache Parquet, ETL design, schema design, data validation, OHLCV pipelines

Docker, Git, GitHub Actions, CI/CD, Linux/Ubuntu, Azure Container Apps, Azure Container Registry, Azure Functions, Azure Storage, Bicep, Terraform, Microsoft Foundry

RabbitMQ, REST integrations, WebSocket feeds, provider adapters, API abstraction layers

MetaTrader 5, cTrader, MatchTrader, trading account operations, market data workflows, platform administration, execution investigation

LangChain, LangGraph, LangSmith, CrewAI, Voiceflow, Zendesk AI, Essel AI, Crisp, RAG systems, agent evaluation workflows, prompt/system-instruction design

A cryptocurrency market data platform for ingesting, validating, storing, querying, and serving OHLCV and funding-rate data.

Core areas:

• Provider abstraction
• Exchange API normalization
• Typed market data records
• Parquet storage
• DuckDB-backed querying
• CLI + REST API access
• Azure Blob Storage support
• Replay-ready data workflows

I am most interested in projects involving:

• backend engineering
• AI platform engineering
• agentic system evaluation
• trading systems
• market data infrastructure
• automation engineering
• data-intensive services
• cloud-backed platform work
• reliability-focused software design

The common thread is building systems where correctness, observability, and operational usefulness matter.