#data-science
Every summary, chronological. Filter by category, tag, or source from the rail.
Building Resilient Web Data Infrastructure for AI
AI systems require live, reliable data pipelines. Success in this space is not about building once, but maintaining an 'adapt forever' architecture that handles extreme scale, latency, and anti-bot measures.
AI EngineerBuilding Production AI: The Data Science & AI Loop
Production-ready AI systems rely on a continuous feedback loop where robust data science pipelines (ETL, governance) feed AI models, and AI, in turn, generates synthetic data to improve those same pipelines.
IBM TechnologyNL2SHACL-Bench: Evaluating LLM Performance on SHACL Generation
NL2SHACL-Bench provides a standardized benchmark suite to evaluate how effectively Large Language Models can translate natural language requirements into SHACL (Shapes Constraint Language) for RDF data validation.
Global AI Trends: From Information Seeking to Task Execution
New data from OpenAI Signals reveals that ChatGPT usage is shifting from exploratory 'asking' to productive 'doing,' particularly in professional settings, with rapid adoption growth in Latin America, Africa, and among users over 35.
The Missing Data Layer in AI Systems
Current AI architectures lack a dedicated, standardized data layer, leading to fragmented pipelines; the proposed solution involves a unified abstraction for data management that bridges the gap between raw storage and model inference.
Scaling AI Weather Forecasting: The WindBorne Strategy
WindBorne Systems raised $37M to scale its proprietary weather-sensing balloon network and AI forecasting models, aiming to bridge the gap between high-fidelity data and commercial business decision-making.
Scaling Human Feedback for AI Model Evaluation
DesignArena, a platform for crowdsourced human evaluation of generative AI, has raised $7.9M to provide frontier labs with high-quality preference data, currently generating $60M in ARR.
AlphaSchema: Semantic Frameworks for LLM-Driven Alpha Mining
AlphaSchema introduces a structured semantic framework to improve how LLMs generate and evaluate quantitative trading signals (alphas), moving beyond unstructured prompt engineering to systematic search spaces.
UrbanDS: Graph-Guided Multi-Agent Systems for Urban Data
UrbanDS improves LLM performance on complex urban data tasks by using a graph-guided multi-agent architecture that structures reasoning and data retrieval.
ClinLens: Long-Horizon Coding Agents for Clinical Data Science
ClinLens is an AI agent framework designed to handle the complexities of longitudinal, multimodal clinical data by automating long-horizon coding tasks in data science workflows.
Data Quality as a Compute Multiplier
Data quality is the most underinvested lever in model training. By curating for signal-per-token rather than raw volume, builders can achieve frontier-level performance with significantly less compute, effectively bending scaling laws.
AI EngineerData Curation Strategies for Post-Training LLMs and Agents
Reliability in autonomous agents is achieved through disciplined data and environment curation rather than just compute, utilizing techniques like multi-answer sampling and targeted SFT.
Building Verifiable AI Benchmarks for Biology
To make AI reliable for biological research, we must move beyond Q&A models and build verifiable, task-based benchmarks that force models to reason through raw experimental data, not just memorize scientific literature.
Unified Semantic Modeling for Large-Scale Job Understanding
LinkedIn's framework addresses the challenge of large-scale job understanding by implementing a unified semantic model that maps diverse, unstructured job data into a standardized, machine-readable format.
LLMs vs. Corpora for Specialized Terminology Extraction
While LLMs offer a flexible alternative to traditional corpus-based methods for extracting specialized terminology, they remain prone to hallucinations and lack the verifiable grounding of static corpora, making them best suited as assistants rather than replacements.
Right-sizing Cloud Workloads with Conformal Prediction
The RSR framework uses conformal prediction to provide statistically rigorous, uncertainty-aware resource recommendations for virtual machines, balancing cost-efficiency with performance guarantees.
Grounding AI in Outcomes: Why Context Isn't Experience
Off-the-shelf LLMs suffer from the 'fluent bluff'—they provide confident but often harmful financial advice because they lack real-world experience. The solution is grounding models in proprietary state-action-outcome data.
AI EngineerSchema-Aware Localisation (SAL) for NL2SQL Reliability
Schema-Aware Localisation (SAL) improves NL2SQL accuracy by grounding natural language queries directly against database schemas in real-time, effectively mitigating hallucinations and invalid SQL generation.
Manufacturing Physical AI Data: Beyond Simple Video Annotation
Physical AI models face a critical data scarcity bottleneck. Companies like Encord are moving beyond passive video collection to 'manufacturing' high-fidelity training data using brain-wave sensors, EMG arm sensors, and dense physical annotations.
The State of Data Markets: Moving Beyond Contrived Benchmarks
Data quality is the primary bottleneck for AI expertise. Success requires moving from 'contrived' type-2 data to 'process-based' type-1 data, while building infrastructure that decouples enterprise workflows from specific foundation models.
AI EngineerEvals-Driven Development for High-Stakes Mental Health AI
SonderMind builds safe mental health AI by replacing generic model guardrails with a modular, clinician-led evaluation loop that treats clinical judgment as code.
AI EngineerFineServe: Analyzing Global LLM Serving Workloads
FineServe provides a comprehensive, fine-grained dataset of real-world LLM serving workloads, revealing critical patterns in request arrival, token distribution, and system utilization that challenge existing assumptions in infrastructure design.
Moving from Multi-Agent Pipelines to Knowledge-Graph Control Planes
Complex multi-agent systems often fail due to context loss and fragmented reasoning. The solution is to use deterministic pipelines for data processing, a single agent for end-to-end reasoning, and a knowledge graph as a control plane to bound agent exploration.
Designing Robust RAG Systems for Complex and Contradictory Data
RAG systems often fail not due to hallucinations, but because they are built on messy, contradictory, or outdated data without proper architectural guardrails to handle ambiguity.
IBM TechnologyClosing the Loop Between Model Evaluation and Data Intervention
By introducing 'capability slices'—groups of evaluation samples categorized by task and operation—engineers can transform benchmark failures into precise, actionable data interventions rather than relying on intuition.
AI-Driven Multi-Document Correlation for Financial Compliance
Moving from isolated document validation to cross-document intelligence using graph-based entity correlation and probabilistic risk modeling significantly improves fraud detection and reduces false positives in enterprise compliance.
AI EngineerMastering Probability Distributions for Machine Learning
Probability distributions are maps of data behavior. Understanding them allows you to select better models, engineer features effectively, and quantify uncertainty in production pipelines.
Why R-Squared Misleads and How to Properly Evaluate Regression
R-squared measures explained variance but ignores model complexity and outliers. To truly understand model performance, you must use a suite of metrics—MAE, MSE, RMSE, and Adjusted R-squared—to identify where your model fails and why.
Refactoring Pandas Workflows with .pipe()
The .pipe() method in Pandas enables cleaner, more readable ETL pipelines by chaining custom functions, reducing boilerplate code and improving maintainability compared to nested or sequential assignments.
Improving Uncertainty Estimation for Classifier Performance
Standard confidence interval methods often fail for small datasets or high-performance models; using Agresti-Coull, Wilson, or regularized bootstrap methods significantly improves accuracy.
Showing 30 of 131