Senior Compiler Engineer (Python → LLVM) for Algorithmic Trading Strategy Engine

AxonQuant Ltd (UK) is building a cloud platform that lets independent traders write, test and run automated FX trading strategies in Python. We are looking for a senior compiler engineer to own the core of the product: pyaxq. WHAT PYAXQ IS pyaxq takes a trader's strategy written in a restricted, typed subset of Python, proves it safe, and compiles it to native code that behaves identically in backtest and live trading. Pipeline: strategy.py → AST → subset check → type inference → Z3 safety proof → IR → LLVM → native .so Guarantees built into the language: - No look-ahead: a strategy can never read data after the current bar - Deterministic: seeded randomness only; no file, network or plotting access - Exact money maths: fixed-point arithmetic (int64, 8 decimals) for prices, pips and P&L - Compiled speed: native code, no interpreter in the hot loop WHAT YOU WILL DO - Define and document the Python subset and the trader-facing SDK (aq.Strategy, aq.Indicator, on_bar, on_tick, order and position methods, aq.Param) - Build the compiler front end: parsing, subset checking, type inference - Use Z3 to prove strategies free of divide-by-zero, out-of-range access and oversized orders - Generate native libraries through LLVM - Design and version a stable C ABI (axq.h) used by the backtester and live engine, with reference hosts in C, Python and Rust - Build compiled, deterministic versions of popular libraries in phases: core NumPy and pandas, the most-used TA-Lib indicators first, then wider coverage (pandas-ta, QuantStats metrics, ML model inference) - Integrate with a nautilus_trader-based backtesting and execution engine (market, limit, stop, SL/TP orders) - Build a conformance test suite and publish benchmarks against plain Python FIRST MILESTONES 1. Language spec v1, axq.h v1 and compiler front end 2. A simple SMA crossover strategy compiled and backtested end to end on real EURUSD data 3. Z3 verification, LLVM code generation and core library support, with a published benchmark 4. Engine integration with deterministic backtest/live replay, and a pip-installable SDK REQUIRED - Built at least one real compiler, interpreter or language toolchain - Hands-on LLVM experience (IR generation, AOT compilation) - Expert C; strong Rust or C++ - Deep knowledge of Python's AST and semantics - Practical experience with SMT solvers (Z3) or other formal methods - Strong grasp of floating-point versus fixed-point arithmetic and reproducibility NICE TO HAVE - Trading systems, backtesting engines or market data experience (FX ideal) - nautilus_trader, Backtrader or Vectorbt - Numba, Cython, mypyc or other Python-to-native work - ONNX / ML inference in native environments HOW WE WORK - Plan posted before each task; all code in our company GitHub, linked to Jira tickets - Every deliverable reviewed by our technical reviewer - Written updates every two days, a weekly plan and a short weekly demo of working code - Technical discussion in a shared Slack channel - All IP is assigned to AxonQuant Ltd; an NDA is required before detailed specs are shared TO APPLY Please answer the screening questions and include a link to a compiler, language or runtime you built or contributed to substantially, with a note on your part in it. Generic proposals will not be considered.