Low-latency Java systems
Multi-threaded, message-driven services built with lock-free concurrency, pre-allocated buffers and GC tuning, for sub-millisecond processing.
What I do
Engineering for systems where speed, correctness and uptime all matter.
Connectivity and integration
FIX sessions, venue integration, REST APIs and message-driven microservices that keep data moving reliably between systems.
Monitoring and resilience
Structured logs, health dashboards, anomaly detection and circuit breakers, so issues are found in minutes and systems recover safely.
Quantitative analytics
Python tools for backtesting, market microstructure and fixed income analytics, built on NumPy, Pandas and kdb+.
Research to production
Taking models and analysis out of the notebook and into tested, deployed code, with CI/CD and automated tests that make every release safer.
AI and NLP workflows
LLM and NLP pipelines that turn unstructured text, like chat messages, into structured data people can act on.
The stack
The languages, systems and tools I work with every day.
Languages
- Java
- Python
- C++
- SQL
- R
- q/kdb+
Systems and data
- FIX protocol
- Event-driven microservices
- kdb+ time-series
- Oracle
- TimesTen
- REST APIs
- Docker
- AWS S3
- Linux/Unix
- TeamCity CI/CD
Analytics and AI
- NumPy
- Pandas
- SciPy
- LLMs
- NLP
- Prompt engineering
Selected work
Built at work and on my own time. Internal details left out on purpose.
Sub-millisecond RFQ processing, from request to fill
A multi-threaded, message-driven Java engine that runs the full order lifecycle for requests for quote across credit and muni instruments, at sub-millisecond latency per message.
- RFQ arrives from a venue
- Validated, queued and processed
- Quote sent, fills handled
- Full lifecycle, sub-millisecond
- Java
- FIX protocol
- Multi-threading
- Lock-free concurrency
- Linux
RFQGen: trader chat in, structured RFQ out
An AI solution that reads unstructured trader messages, extracts the RFQ details and drafts the ticket, with GenAI trade suggestions on top. 3rd place in its panel at the Barclays GenAI Hackathon.
- Trader sends a chat message
- LLM extracts the RFQ details
- Ticket drafted with trade suggestions
- RFQ ready, no retyping
- Python
- LLMs
- NLP
- Prompt engineering
Backtest routing rules before they go live
A parameterized Python framework that replays historical RFQ data against different order-routing rules and scores each one on hit rate, adverse selection and P&L.
- Pick routing rule variants
- Replay historical RFQ data
- Score hit rate, adverse selection
- Compare P&L per configuration
- Python
- Pandas
Live diagnostics that find order-flow problems in under a minute
Structured latency logs, FIX session health dashboards and market data gap detection that turn "something feels slow" into a clear cause, fast.
- Latency spike or data gap
- Health checks flag it
- Fail-safes reconnect or degrade gracefully
- Diagnosed in under a minute
- Java
- FIX protocol
- Structured logging
- Anomaly detection
- Circuit breakers
Reading the market from tick and quote data
A Python and kdb+ toolkit that turns raw tick and quote data into spread decomposition, order flow imbalance, VWAP deviation and short-horizon price impact.
- Raw tick and quote data
- Clean and align time series
- Compute spreads, imbalance, VWAP
- Clear view of trading costs
- Python
- Pandas
- kdb+/q
Deployments 5× faster, with 40% less QA effort
A rebuilt CI/CD pipeline and automated test suite that made releases faster and safer for a trading platform.
- Code change committed
- Pipeline builds and tests it
- Regression suite checks behavior
- Deployed 5× faster
- TeamCity
- CI/CD
- Java
- Automated regression testing
How I work
Start with the real problem
I work directly with the people who feel it, like traders, analysts or users, and turn symptoms such as stale prices or latency spikes into a precise technical cause.
Build it to be measured
Clean, tested code with latency logs and health checks built in from day one, so performance is a number, not a guess.
Ship it and keep it healthy
Automated tests and CI/CD for safe releases, then monitoring and fail-safes so the system keeps running when something goes wrong.
Now
What I'm focused on right now.
- March 2026
Promoted to AVP, Software Engineer at Barclays Investment Bank.
- 2026
Building Python tools for backtesting, market microstructure and fixed income analytics.
- 2026
Exploring how LLMs can support research and make financial data easier to interpret.
Building something that has to be fast and reliable?
Send me a message. I'm always happy to talk engineering.