Advanced Data Scientist

Key Responsibilities • Mathematical Formulation: Translate ambiguous business problems into mathematically sound framework objectives and optimisation targets. • Production-Grade Engineering: Write clean, modular, and maintainable code using production-level design patterns to scale mathematical models. • Big Data Processing: Design and manage scalable data pipelines to process massive datasets efficiently for model training and inference. • Deep Learning & Vision Development: Build, train, and fine-tune complex neural networks across text, audio, and visual modalities. • Cloud Deployment: Architect and deploy models to cloud environments, leveraging distributed computing and robust cloud infrastructure. Required Technical Skills & Competencies 1. Tooling, Libraries & Software Engineering • Core Language: Advanced proficiency in Python with a strict adherence to Object-Oriented Programming (OOP) principles, clean coding standards, and design patterns. • Machine Learning Libraries: Advanced proficiency in scikit-learn (sklearn) for data preprocessing, feature engineering, and baseline modelling. • Deep Learning Frameworks: Core expertise in PyTorch (preferred) or TensorFlow for building, customizing, and training deep neural networks from scratch. • Big Data Ecosystem: Experience with Apache Spark (PySpark) and the Hadoop Ecosystem (HDFS, Hive, MapReduce) for handling, transforming, and querying large-scale distributed datasets. • Cloud Architecture: Experience building and deploying scalable machine learning applications on major cloud platforms (AWS, Azure, or GCP). 2. Core Mathematics & First-Principles ML • Foundational Math: Solid foundation in Linear Algebra (eigenvalues, SVD, matrix decompositions), Multivariable Calculus (partial derivatives, gradients, Jacobians), and Probability Theory (Bayesian inference, probability distributions, expectation maximization). • Machine Learning: In-depth understanding of standard Machine Learning algorithms (Trees, Boosting, SVMs, GMMs) with the ability to explain the underlying loss functions and optimizations mathematically. • Deep Foundations: Thorough understanding of Multi-Layer Perceptrons (MLPs), mathematical derivation of backpropagation, hyperparameter initialization strategies (Xavier, He), optimization variants (Adam, RMSProp), and advanced regularization techniques (L1/L2, Dropout, Batch Normalization). 3. Advanced Natural Language Processing (NLP) • Sequential Networks: Hands-on experience with sequence modeling, including Word Embeddings (Word2Vec, FastText), RNNs, LSTMs, and GRUs. • Transformer Ecosystem: Deep structural knowledge of the Transformer architecture (Self-Attention math, Multi-Head mechanisms). • Pre-trained NLP Models: Experience implementing and fine-tuning encoder-only (BERT, RoBERTa) and decoder-only (GPT series) architectures. 4. Computer Vision (CV) & Document AI • Spatial Networks: Deep understanding of Convolutional Neural Networks (CNNs), feature map mathematics, pooling operations, and advanced CV backbones. • OCR & Document Processing: Proven track record building or customizing Optical Character Recognition (OCR) systems for complex text extraction pipelines. • Vision Transformers: Familiarity with the adaptation of attention mechanics to visual tasks (ViTs, Swin Transformers). Education & Experience • Education: Bachelor’s, Master's, or Ph.D. in a highly quantitative field (Mathematics, Statistics, Econometrics, Computer Science, Physics, or Operations Research). • Experience: 4 to 6 years of industry experience working as a Data Scientist or Machine Learning Engineer with a portfolio of complex multimodal projects.