Senior Data Science Engineer

Trinetix; Berlin; Vor Ort; EUR 80.000 - 120.000; Vollzeit

Benefits dieser Stelle

Professional training

Medical insurance

Mental health support

Flexible working hours

Inclusive culture

Zusammenfassung

Trinetix seeks a Data Science Engineer to craft advanced forecasting and optimization solutions. You will design models like XGBoost, BSTS, and TFT, with hierarchical reconciliation for 545+ locations, and build MILP/min-cost flow pipelines for cascading optimization.

The role includes document intelligence using Textract and LayoutLMv3, RAG pipelines, and deployment on AWS SageMaker with MLOps, SHAP/LIME explainability, and model monitoring.

Qualifikationen

• Experience with Python and ML frameworks (PyTorch or TensorFlow).
• Strong knowledge of time-series forecasting tools (BSTS, Prophet, TFT).
• Proficiency in optimization (MILP, min-cost flow) and related solvers (PuLP, OR-Tools).
• Hands-on experience with SageMaker and MLOps features (Model Monitor, Feature Store, Pipelines).
• Familiarity with document AI tools (Textract, LayoutLMv3) and RAG pipelines.

Aufgaben

• Build demand forecasting models (XGBoost, BSTS, TFT) with hierarchical reconciliation across many locations.
• Develop optimization pipelines using MILP/Min-Cost Flow and solver integrations.
• Implement document intelligence pipelines (Textract, LayoutLMv3) and RAG with Bedrock.
• Deploy and monitor models on SageMaker with MLOps features.

Kenntnisse

Python

PyTorch

TensorFlow

scikit-learn

XGBoost

LightGBM

pandas

NumPy

BSTS

Prophet

Temporal Fusion Transformer

MinT reconciliation

Linear Programming

MILP

PuLP

OR-Tools

SageMaker

Textract

LayoutLMv3

RAG

Bedrock

OpenSearch

GNN

Deep RL

Survival analysis

Cox Proportional Hazards

XGBoost-Survival

attention models

SHAP

LIME

Captum

MLflow

Tools

PuLP

OR-Tools

Jobbeschreibung

We are seeking a skilled Data Science Engineer.

Must-Have Requirements

• Programming & ML Frameworks: Python; PyTorch or TensorFlow; scikit-learn; XGBoost or LightGBM; pandas; NumPy
• Time Series & Forecasting: BSTS; Prophet; Temporal Fusion Transformer (TFT); hierarchical forecasting with MinT reconciliation
• Optimization: Linear Programming and MILP using tools such as PuLP and OR-Tools; constraint satisfaction; min-cost flow optimization
• AWS ML Stack: Amazon SageMaker (Training Jobs, Endpoints, Model Monitor, Clarify, Feature Store, Pipelines)

Nice-to-have

• NLP & Document AI: Amazon Textract; LayoutLMv3; Retrieval-Augmented Generation (RAG) pipelines; Amazon Bedrock (Claude); OpenSearch vector databases
• Advanced Machine Learning: Graph Neural Networks (GNNs); Deep Reinforcement Learning; Survival Analysis (Cox Proportional Hazards, XGBoost-Survival); attention-based models
• Explainability & MLOps: SHAP, LIME, Captum; MLflow; A/B testing; champion/challenger frameworks; model and data drift detection

Core Responsibilities

• Build demand forecasting models (XGBoost, BSTS, Temporal Fusion Transformer) with hierarchical reconciliation across 545+ locations
• Develop cascading optimization using MILP/Min-Cost Flow solvers (PuLP, OR-Tools, Gurobi) and Hybrid ML+Optimization pipelines
• Implement document intelligence pipeline: Textract + LayoutLMv3 for document extraction, RAG with Bedrock (Claude) for semantic reasoning
• Deploy models on SageMaker with MLOps (Model Monitor, Feature Store, Pipelines); implement SHAP/LIME explainability Models You'll BuildDemand Forecasting: Gradient-boosted models (XGBoost), Bayesian Structural Time Series (BSTS), and Temporal Fusion Transformers (TFT), including hierarchical reconciliation Cascading Optimization: Mixed-Integer Linear Programming (MILP) and Min-Cost Flow models, evolving to hybrid ML + solver approaches and advanced Graph Neural Network (GNN) and Deep Reinforcement Learning (DRL) solutions Document Intelligence: Automated document extraction using Amazon Textract and LayoutLMv3, advancing to Retrieval-Augmented Generation (RAG) pipelines with Amazon Bedrock and Vision-Language Models Survival & Out-of-Spec Prediction: Kaplan-Meier estimators, Cox Proportional Hazards models, and XGBoost-Survival techniques

What we offer

• Continuous learning and career growth opportunities
• Professional training and English/Spanish language classes
• Comprehensive medical insurance
• Mental health support
• Specialized benefits program with compensation for fitness activities, hobbies, pet care, and more
• Flexible working hours
• Inclusive and supportive culture

About Us

Established in 2011, Trinetix is a dynamic tech service provider supporting enterprise clients around the world.

Headquartered in Nashville, Tennessee, we have a global team of over 1,000 professionals and delivery centers across Europe, the United States, and Argentina.

We partner with leading global brands, delivering innovative digital solutions across Fintech, Professional Services, Logistics, Healthcare, and Agriculture.Our operations are driven by a strong business vision, a people-first culture, and a commitment to responsible growth.

We actively give back to the community through various CSR activities and adhere to international principles for sustainable development and business ethics.

To learn more about how we collect, process, and store your personal data, please review our Privacy Notice: https://www.trinetix.com/corporate-policies/privacy-notice