Python Developer

About the Project We operate a proprietary Python research platform for systematic trading at a private investment firm. The codebase is mature, well-tested, and disciplined: ML model training pipelines, a backtesting engine built on VectorBT Pro, a versioned model-release system with strict reproducibility guarantees, and a reporting layer that feeds downstream portfolio processes. We're looking for a careful, detail-oriented Python developer to maintain and extend this platform full-time. This is not greenfield development — the architecture and validation discipline are established. Your job is to keep the system running cleanly, execute release cycles, and deliver well-scoped improvements without breaking reproducibility. What You'll Do - Maintain the data pipeline: MongoDB-backed time-series data, ingestion jobs, coverage and integrity checks, data-drift monitoring - Maintain and extend the research tooling: a CLI-driven workflow (Click) covering model retraining, hyperparameter search (Optuna), backtest validation, and a hash-manifested release process - Run and verify releases: execute the release checklist (validate → replay → backtest → tag), and investigate any reproduction mismatch down to root cause - Keep the test suite green (~300 pytest tests) and add tests alongside every change - Generate reporting artifacts: equity curves, performance analytics, charts, PDF tearsheets - Well-scoped feature work from written specs: new CLI commands, diagnostics, report variants Required Skills - Strong Python (3.12, type hints, dataclasses) with pandas/NumPy fluency — most of the work is time-series manipulation where subtle alignment errors have real consequences - Solid testing habits: pytest, regression testing, verifying changes against recorded baselines - Git discipline: clean commits, tags, release workflows - Experience with scikit-learn / LightGBM (training, persistence, deterministic re-inference) - MongoDB basics (pymongo, aggregation pipelines) - Comfort with CLI tooling (Click) and uv-managed environments