1 Datascie Jobs

Required Skills & Experience Experience Level 5-8 years of professional experience in Data Science , ML / DL Architecturing , Generative AI engineering or AI/ML development roles Proven track record of deploying and maintaining production ML systems at scale Experience in digital marketing, customer service, contact centers, or related industries strongly preferred Technical Expertise Programming & Core Libraries Python Ecosystem Mastery (Required) Advanced Python 3.9+ proficiency with deep understanding of object-oriented programming, decorators, context managers, and asyncio for concurrent processing Data Manipulation & Analysis: Pandas 2.0+: Advanced operations including multi-indexing, groupby operations, window functions, categorical data handling, and performance optimization techniques NumPy 1.24+: Vectorized operations, broadcasting, advanced indexing, custom ufuncs, and memory-efficient array operations Polars: Next-generation DataFrame library for faster data processing and memory efficiency Big Data Processing: PySpark 3.4+: DataFrame API, Spark SQL, MLlib, structured streaming, and cluster optimization Dask: Parallel computing, delayed operations, distributed arrays, and scaling pandas workflows Scientific Computing: SciPy: Statistical functions, optimization algorithms, signal processing, and numerical integration Statsmodels: Advanced statistical modeling, time series analysis (ARIMA, GARCH, VAR), and econometric analysis Scikit-learn 1.3+: Custom transformers, pipeline optimization, advanced cross-validation strategies, and model interpretation tools Machine Learning Frameworks (Expert Level) Deep Learning Platforms: PyTorch 2.0+: Dynamic computational graphs, autograd, custom loss functions, distributed training, and TorchScript optimization TensorFlow 2.13+: Eager execution, tf.function optimization, custom layers, TensorBoard integration, and TensorFlow Serving Keras 3.0: High-level API mastery, custom metrics/callbacks, transfer learning, and model subclassing Gradient Boosting Libraries: XGBoost 2.0+: Advanced hyperparameter tuning, custom objective functions, early stopping, and SHAP integration LightGBM 4.0+: Categorical feature handling, dart boosting, and GPU acceleration CatBoost: Handling high-cardinality categorical features and automatic feature engineering Specialized ML Libraries: Optuna/Hyperopt: Bayesian optimization for hyperparameter tuning SHAP/LIME: Model interpretability and explainable AI Imbalanced-learn: Handling class imbalance with SMOTE, ADASYN, and ensemble methods Generative AI & LLM Technologies Large Language Model Platforms (Production Experience Required) OpenAI Ecosystem: GPT-4/GPT-4 Turbo: Advanced prompting strategies, function calling, vision capabilities, and fine-tuning with GPT-3.5 OpenAI API: Rate limiting, error handling, cost optimization, and embedding API usage Whisper: Speech-to-text integration for call center voice analysis DALL-E: Image generation for automated content creation Google AI Platform: Gemini Pro/Ultra: Multimodal capabilities, context length optimization, and enterprise integration Vertex AI Model Garden: Deployment and management of foundation models Meta AI Models: Llama 2/3: Open-source deployment, custom fine-tuning, and performance optimization Code Llama: Automated code generation and review processes Anthropic Claude: Constitutional AI principles, safety considerations, and enterprise deployment Open Source Alternatives: Mistral AI, Falcon, MPT, and Alpaca models for cost-effective solutions GenAI Development Frameworks & Tools LangChain 0.1+: Advanced chain composition, memory management, and custom tool creation Agent frameworks for autonomous task execution Integration with external APIs and databases Streaming responses and async processing LangGraph : Stateful multi-agent workflow orchestration and complex reasoning chains Cyclic graph construction for iterative problem-solving and self-correction loops State management for long-running conversational agents Human-in-the-loop integration for escalation and approval workflows Parallel execution and conditional branching for complex decision trees Graph visualization and debugging for workflow optimization LangSmith : Production LLM monitoring, logging, and performance analytics A/B testing frameworks for prompt optimization and model comparison Dataset creation and management for fine-tuning and evaluation Automated evaluation pipelines with custom metrics and human feedback integration Error tracking and debugging for complex multi-step LLM workflows Cost monitoring and optimization across different LLM providers Annotation tools for creating high-quality training datasets Production deployment monitoring with alerting and anomaly detection Vector Databases & Semantic Search Weaviate: GraphQL APIs, custom vectorization, and multi-modal search Advanced LLM Techniques Fine-tuning Methodologies: LoRA (Low-Rank Adaptation): Memory-efficient fine-tuning for domain adaptation QLoRA: Quantized LoRA for reduced memory footprint Instruction Tuning: Creating custom instruction datasets and training procedures RLHF (Reinforcement Learning from Human Feedback): Implementing reward models and PPO training Prompt Engineering: Advanced prompting techniques: Chain-of-thought, few-shot learning, and constitutional AI Prompt optimization: Systematic prompt testing and A/B testing frameworks Dynamic prompting: Context-aware prompt generation and adaptive prompting strategies RAG (Retrieval-Augmented Generation): Advanced RAG architectures: Multi-hop reasoning, query decomposition, and re-ranking strategies Hybrid search: Combining dense and sparse retrieval methods Context optimization: Document chunking strategies, overlap management, and context compression Model Optimization: Quantization: INT8/INT4 quantization for inference optimization Distillation: Knowledge distillation from large to smaller models Pruning: Structured and unstructured pruning techniques Data Science Methodologies & Statistical Analysis Advanced Statistical Methods Hypothesis Testing & Experimental Design: A/B Testing: Power analysis, sample size calculation, multiple testing correction (Bonferroni, FDR) Bayesian Statistics: Prior specification, MCMC sampling, hierarchical modeling with PyMC or Stan Causal Inference: Difference-in-differences, instrumental variables, propensity score matching, and causal DAGs Survival Analysis: Cox regression, Kaplan-Meier estimation, and time-to-event modeling Multivariate Analysis: Dimensionality Reduction: PCA, t-SNE, UMAP, Factor Analysis, and manifold learning techniques Clustering: K-means++, DBSCAN, hierarchical clustering, and mixture models Classification & Regression: Logistic regression, SVM, ensemble methods, and neural networks Time Series Analysis: Classical Methods: ARIMA, SARIMA, Exponential Smoothing, and seasonal decomposition Modern Techniques: Prophet, Neural Prophet, and Transformer-based forecasting Multivariate Time Series: VAR models, Granger causality, and state-space models Feature Engineering & Data Preprocessing Advanced Feature Engineering: Automated Feature Engineering: Featuretools, AutoFeat, and genetic programming approaches Feature Selection: Recursive feature elimination, mutual information, and stability selection Categorical Encoding: Target encoding, entity embeddings, and hash encoding Text Feature Engineering: TF-IDF variants, word embeddings, and n-gram analysis Data Quality & Validation: Data Profiling: Automated data quality assessment and anomaly detection Missing Data Handling: Multiple imputation, matrix completion, and missingness pattern analysis Outlier Detection: Isolation Forest, Local Outlier Factor, and robust statistical methods Data Drift Detection: Population stability index, KL divergence, and distribution monitoring Model Development & Validation Cross-Validation Strategies: Time Series CV: Walk-forward validation, blocked cross-validation, and gap-based splitting Stratified CV: Maintaining class balance and group-based splitting Nested CV: Unbiased hyperparameter optimization and model selection Model Interpretability: Global Interpretation: Feature importance, partial dependence plots, and SHAP summary plots Local Interpretation: LIME, SHAP values, and counterfactual explanations Model-Agnostic Methods: Permutation importance and accumulated local effects Performance Evaluation: Classification Metrics: ROC-AUC, PR-AUC, F1-score variants, and calibration assessment Regression Metrics: MAPE, SMAPE, directional accuracy, and residual analysis Business Metrics: Customer lifetime value, churn prediction accuracy, and revenue impact assessment Advanced Statistical Modeling & Machine Learning Contact Center Analytics & Voice Intelligence Speech Analytics & NLP: Conversation Intelligence: Automatic topic detection, intent classification, and sentiment progression Voice Emotion Recognition: Acoustic feature extraction and emotion classification models Speaker Analytics: Voice biometrics, accent detection, and speech quality assessment Workforce Analytics: Agent Scheduling Optimization: Integer programming for shift scheduling with constraints Performance Prediction: Predicting agent success during hiring and ongoing performance Training Effectiveness: Measuring training ROI and skill development tracking Customer Journey Analytics: Omnichannel Attribution: Cross-channel customer journey mapping and touchpoint analysis Intent Prediction: Predicting customer intent from early interaction signals Effort Prediction: Customer effort score prediction and friction point identification Required Education Bachelor's or Master's degree in Computer Science, Software Engineering, Data Science, Mathematics, or related technical field Advanced degree preferred for senior-level positions Industry Experience Domain Expertise (Preferred) Call center and customer service industry experience with understanding of contact center operations Telecommunications or SaaS industry background Customer experience (CX) technology and analytics Real-time systems development for customer-facing applications Regulatory compliance experience in data-sensitive industries