Machine Learning

Experience and skills you need to join us: 3+ years of practical experience delivering ML or AI features to production (not just notebooks). Hands-on experience building LLM-based solutions: retrieval, context injection, grounding answers, prompt design. Experience with RAG pipelines: document chunking, embedding generation, retrievalquality tuning, hallucination control. Strong Python engineering skills (clean code, reproducible pipelines, packaging services). Solid understanding of classic ML / DL (pandas, numpy, scikit-learn, Keras or PyTorch) — we still do forecasting, anomaly detection, predictive maintenance. Comfortable working in containers (Docker) and basic CI/CD. Ability to work with noisy real-world data: gaps, outliers, inconsistent tags, messy sensor histories. Able to communicate results and limitations clearly to non-ML stakeholders It would be great if you also have: Experience deploying LLM / RAG stacks fully on-prem (air-gapped / no public API calls). Background in power generation / heavy industry (boilers, turbines, condensers, outages, performance KPIs). Experience creating internal tools / dashboards for engineers or operators (Dash, Plotly, lightweight React). Familiarity with NoSQL or time-series data from industrial sensors. Exposure to AWS or similar cloud mainly for experimentation, not for runtime. Experience mentoring teammates or driving architecture decisions. Offer descriptionWe build AI assistants and optimization models for real industrial assets — turbines, boilers, condensers, cooling systems. The systems you build will sit in environments where public cloud is not allowed, so you’ll work on serious on-prem LLM / RAG problems: secure retrieval, auditability, explainability, repeatability.You won’t be a “ticket closer.” You’ll shape how our on-prem LLM stack works: retrieval strategy, evaluation loop, model selection, UX for operators. You’ll also see direct business impact in MW saved, downtime avoided and CO₂ reduced. If you like applied LLM work with real consequences (not just demos), this is that role. Benefits 100% remote and flexible schedule. Direct influence on architecture and product direction. You ship systems that will actually be used on-site in critical infrastructure. Low-politics environment, high ownership. Support for growth (LLM eval, retrieval quality, industrial analytics, energy domain training) Equipment / environment High-performance local hardware for working with LLMs and ML workloads. Access to domain experts who actually run power plants. Access to historical telemetry, maintenance logs, outage reports, efficiency KPIs ,[Build, tune and maintain on-prem LLM solutions with RAG (diagnostics assistants, operator copilots, root-cause explainers)., Design retrieval pipelines and knowledge bases: ingestion, chunking, embeddings, metadata, access control., Control quality: grounding, hallucination reduction, evaluation of answer accuracy and usefulness for real operators., Integrate LLM outputs into production workflows (dashboards, alerts, reports for plant engineers)., Develop ML / DL models for predictive maintenance, efficiency optimization and digital twins (virtual assets of turbines / boilers / condensers)., Own data pipelines: from raw sensor streams and maintenance logs to features ready for inference., Package everything into reliable Python services, ship via Docker, keep it maintainable., Work directly with domain experts (performance engineers, plant ops) to translate real operational problems into AI use cases with measurable impact] Requirements: prompt engineering, Python, RAG, Docker, Git, AI, Clean Code, pandas, NumPy, scikit-learn, Keras, PyTorch, Forecasting, Dash, React, CI/CD, NoSQL, AWS Additionally: Support for growth, Low-politics environment, high ownership.