About
My projects
Systems I’ve built
UberHF 2.0 (low-latency trading)
Trading framework in pure C (not C++) for low-latency market-making and arbitrage. UberHF 1.0 started in 2022; the pure C rewrite, 2.0, started in 2023.
Highlights:
- Micro-app architecture
- In-memory quotes and datafeed
- Shared-memory events and RPC: no system calls on the hot path
- 67k lines of C, including 35k of unit tests
- Smart execution and algorithmic orders
- Stocks, futures and options
- Custom FAST parser, 5–7x faster than mFAST
- FIX (SBE) order management
MicroStrategy Framework
Low-frequency Python execution framework for model trading, smart order execution and complex positions. In production since 2021.
Highlights:
- Independent position handling
- Internal netting of execution orders to cut transaction costs
- Multi-leg execution: baskets, spreads, pairs
- Multi-market, multi-broker execution
- Smart orders: time-sliced orders, options delta-hedge orders
- Historical backtesting for options, spreads, pairs, futures and other complex positions
- 61k lines of Python, including 45k of unit tests
UberMiner
The analysis engine behind my own trading: signal generation and modelling. In development since 2017.
Highlights:
- OmniFeed: one datafeed for stocks, futures, spreads, pairs and options
- DMlib: alpha generation and analysis library
- Monte-Carlo simulation of portfolio risk
- ML-driven market research
- Indicators in Numba: vectorised, no look-ahead, predictable on bad data. Based on yauber-algo.
- AsyncIO microservices on yauber-executor
- 69k lines of Python and Cython
Projects for clients
Swarm Investments LLC (2023 – 2026)
Sole quant developer and architect on contract. Adapted MicroStrategy for execution and position management and UberMiner for analysis and datafeed. Built the Hive© tool for generating alpha swarms and signals on US futures, FIX execution through Trading Technologies, and the Linux server infrastructure.
TMQR EXO Trading Framework (2018 – 2022)
Lead quant developer and architect. Framework for research, backtesting and running a multi-account fund on futures, options and US stocks. The framework ran $40M+ of capital across multiple accounts.
Highlights:
- Backtesting engine for machine-learning models on futures, stocks and options
- Datafeed and multi-account execution infrastructure
- In-house Jupyter research environment over VPN
- Core in Python (31k lines) and Jupyter notebooks (20k lines)
- Linux server infrastructure
- Execution via Trading Technologies FIX and Interactive Brokers API
Open-source projects
AsyncFIX - AsyncIO FIX Protocol Python Framework
AsyncFIX - AsyncIO FIX Protocol Python Framework
Highlights:
- Pythonic FIX messages
- Full AsyncIO support
- Schema validation
- Tools for unit testing client apps
- NewOrderSingle container and state management
- Session management: journaling, session restore, heartbeats, resets
- FIX Tester for unit tests: schema validation, order management, message exchange
- FIX 4.4
- 100% unit test coverage
UberHF 1.0 Cython (archived)
UberHF Cython (archived/pre-alpha)
First generation of my high-frequency framework, in C, Cython and Python. Archived in favour of the pure C rewrite, UberHF 2.0.
Highlights:
- Microservice architecture
- Binary protocol over ZeroMQ between apps
- FIX friendly
- In-memory shared datafeed
- Smart order execution
YAUBER - Algo Lib
Yet Another Universal Backtesting Engine Release (YAUBER) - Algo Lib
Standalone algorithms for financial time series analysis.
Highlights:
- Numba-compiled, generally faster than equivalent Pandas code
- No look-ahead, consistent when the history start changes, NaN friendly, built-in data validation
- Stable API, 100% unit test coverage
YAUBER - Backtester
Yet Another Universal Backtesting Engine Release (YAUBER) - Backtester
Bare-bones portfolio backtesting engine:
- Asset ranking, basket trading, portfolio rebalancing
- Large universes (2000–3000 US stocks, daily) or small intraday ones (futures or FX, hourly)
- Meta-strategies: portfolios of other strategies
- Capital allocation, costs and margin simulation
YAUBER - Executor
Yet Another Universal Backtesting Engine Release (YAUBER) - Executor
Boilerplate for asyncio Python distributed infrastructure.
- Fully asynchronous
- RabbitMQ and MongoDB
- Messaging and RPC between scripts
- Sample infrastructure
- In production 24/7 since 2019
Cython Dev Tools: toolkit for efficient Cython development
Cython Dev Tools: toolkit for efficient Cython development
Key features
- All development tools in one place
- Built-in debugger, including cdef and C code, on GDB
- Unit testing with coverage reports
- Line profiler
- Run .pyx files by entry-point function
- Annotations and an index for the whole project
- Project boilerplate
- Snippets for low-level C code and debugging
Cex.C - Comprehensively EXtended C Language
Cex.C - Comprehensively EXtended C Language
Standalone, dependency-free, single-header C language extension for GCC/Clang. Adds modern language features while remaining plain C11 (C23 compatible).
Highlights:
- Single header under 20k lines, no dependencies beyond GCC/Clang and libc
- Integrated build system: CEX builds itself, no external build tool needed
- Allocator-based memory management: temporary scopes, arenas, auto-free
- Namespaces for grouping functions and simulating OOP classes
- Error handling with stack tracebacks and assertions
- Built-in unit test runner, code generation and help system
- Cross-platform, multi-architecture, big/little endian
- Type-safe dynamic arrays and hashmaps, string views, wildcard matching
cexstdextension library with JSON support and serialization code generation