Senior Machine Learning Engineer Remote, Warsaw 220 PLN/hour

7bulls; Warszawa; Hybrid; PLN 260,000 - 420,000; Full time

Benefits offered by this job

Private medical care

Sports card / co-funding for athletic

Long-term B2B cooperation

Job summary

7bulls is seeking a Senior Machine Learning Engineer to own R&D in time-series forecasting and set rigorous experimental standards. You will lead technical discussions with clients in English, review team outputs for validation integrity, and choose Pragmatic tool directions such as gradient boosting, neural nets, or time-series foundation models.

The role emphasizes production-grade Python and reproducible environments.

Qualifications

• Must have independently delivered end-to-end forecasting projects on noisy time series (crypto, finance, biosignals, CTR, predictive maintenance).
• Mastery in designing time-series validation schemes (walk-forward, purge/embargo, window selection, regime changes).
• Awareness that high-noise environments can still yield valid wins at 52% accuracy.
• Ability to build quick baselines in days and avoid unnecessary deep models if baselines win.
• Strong foundation in statistics: significance, power, multiple testing, confidence intervals, stationarity.

Responsibilities

• Own R&D: end-to-end forecasting experiments from hypotheses to AI execution and reporting.
• Establish Research Standards: define and enforce experiments with autonomy.
• Technical Peer Review: verify validation, check data leakage, assess model stability.
• Pragmatic Tool Selection: decide when to use gradient boosting, neural nets, or time-series foundation models.
• Client Engagement: lead technical discussions and present findings in English to clients.

Skills

Forecasting time series

Time-series validation

Low SNR analysis

Baseline modeling

Statistical significance

Production Python

Git & reproducible environments

DuckDB

Parquet

Docker

LLM tooling

Polish (native)

English for clients

Tools

DuckDB

Parquet

Docker

venv

Job description

At 7bulls, we build advanced software powered by Artificial Intelligence and AI Agents. We value clean engineering, pragmatism, and real business impact, working in an agile manner without unnecessary ceremonies or corporate bureaucracy. We specialize in forecasting complex, market-driven time series. Having achieved consistent success in the energy sector, we are now aggressively expanding into the cryptocurrency space.

We are seeking a Senior Machine Learning Engineer to join a team composed of seasoned experts with over a decade of experience. We need a true technical peer and sparring partner—an experienced engineer who will take full ownership of their R&D research, set rigorous experimental standards for the entire team, and ensure smooth, direct technical communication with clients.

• Own R&D: Drive time-series forecasting experiments end-to-end—from formulating hypotheses to agentic AI execution, reporting, and defending conclusions.
• Establish Research Standards: Define and enforce experimental methodology across the team (clear hypothesis, proper data split, baseline, stop criteria). You will have complete autonomy in this area.
• Technical Peer Review: Review team outputs—verify validation integrity, catch data leakage, and evaluate model stability across time windows.
• Pragmatic Tool Selection: Decide when to leverage gradient boosting, neural networks, or time-series foundation models (Chronos, TimesFM), and when to abandon a specific direction.
• Client Engagement: Lead technical discussions and present research findings in English directly to clients.

Our perfect match (Must Have)

• Experience in Challenging Domains: Proven track record of independently delivering end-to-end forecasting projects on noisy time series with a low Signal-to-Noise Ratio (SNR) (e.g., crypto, finance, biosignals, CTR, predictive maintenance), including the ability to articulate why and when failed experiments were terminated.
• Advanced Time-Series Validation: Mastery in designing validation schemes tailored to temporal data (walk-forward, purging, embargo, window selection, regime changes) and a deep understanding of why random k-fold cross-validation is fundamentally flawed for time series.
• Deep Understanding of Low SNR: Clear awareness that in high-noise environments, an accuracy of 52% can be a major win, whereas 65%+ almost certainly signals validation error or data leakage.
• Baseline-First Mindset & Execution Speed: Ability to build a quick, simple baseline model in days rather than weeks, and a readiness to abandon complex neural networks if they fail to outperform the baseline.
• Solid Statistical & Analytical Foundation: Hands-on command of statistical significance, test power, multiple testing issues, confidence intervals, and stationarity. Ability to interpret network behavior (learning curves) and spot overfitting immediately.
• Software Engineering & Tech Stack: Production-grade Python, disciplined Git hygiene, reproducible environments (uv, venv, Docker), and daily hands-on experience with DuckDB and Parquet.
• Pragmatic LLM Usage: Leveraging AI tools for pipelines and boilerplate code while maintaining strict, manual verification of split logic and target construction.
• Agile Autonomy & Language Proficiency: Self-directed planning without needing micro-managed Jira tickets, balanced with the self-awareness to consult team experts when needed. Native/fluent Polish for internal teamwork and fluent English for client-facing technical discussions.

Nice to Have

• Domain knowledge of energy markets (RDN, TGE, PSE) or cryptocurrency market mechanics (spot vs. derivatives).
• A cool-headed, critical perspective on academic benchmarks for time-series foundation models.

We Offer

• Long-term, stable B2B cooperation.
• Flexible working hours tailored to your needs.
• Private medical care.
• Sports card / co-funding for athletic activities.
• An individual, flexible, human-centric approach to work and life circumstances.

NIP: 5213218694
KRS: 0000139981
REGON: 015271517