Tokenless

Tokenless

How Revolut Cut AI Coding Costs 90% With Self-Hosted Open Models - David Abitbol, Revolut

Interactive discovery

Explore the topic map

Follow the connections between themes, people, and ideas across the Tokenless archive in an interactive topic modeling map.

Open the map Browse posts

Machine Learning

View All
Judge a Transaction by the Company It Keeps — Somya Rai & MD Sharique, EXL

Judge a Transaction by the Company It Keeps — Somya Rai & MD Sharique, EXL

EXL developed an advanced fraud detection system that augments XGBoost with Graph Neural Network (GNN) embeddings. Traditional tabular models struggle to capture fraud signals residing in transaction neighborhoods. The system's innovative design includes meticulous sampling strategies, disjoint monthly graph construction for leakage control, and a GraphSAGE architecture that aggregates mean, max, and standard deviation of neighbor features to detect subtle fraud patterns. Emphasizing explainability for regulatory compliance and analyst trust, it also features a highly optimized serving architecture achieving 3ms P99 latency, pivoting to edge prediction to reduce cache footprint from 60GB to 2-3GB. This led to a 27-33% relative lift in true positive rates and a projected $65 million incremental fraud capture.

From 15% to 90% GPU Utilization: Fix the Data Pipeline, Not the Model

From 15% to 90% GPU Utilization: Fix the Data Pipeline, Not the Model

Tarun Sunkaraneni from Amazon AGI shows how a multimodal SFT pipeline that spent about 85% of its time waiting on data was brought to roughly 90% GPU utilization. He walks through concurrency with asyncio and Ray actors, prefetching, zero-copy transfer via Ray's object store, and a network-card bottleneck that appears only at scale, and argues that the hardest part of training throughput is often keeping GPUs fed rather than optimizing the model.

Small Models, Big Results: Training a Finance Agent for Under $500 — Charles Dickens, Snorkel AI

Small Models, Big Results: Training a Finance Agent for Under $500 — Charles Dickens, Snorkel AI

Snorkel AI and UC Berkeley's Sky Computing Lab show that a 4B Qwen3 model, trained with reinforcement learning for under $500 using the rLLM framework, beats its 235B sibling on expert-validated financial question answering built from SEC 10-K filings. The talk covers the FinQA dataset construction and verification, common failure modes of frontier models, transfer to multi-table and general tool-use tasks, and ablations showing that simple data and binary rewards outperform more elaborate setups, since the real bottleneck is disciplined tool use rather than reasoning depth.

Artificial Intelligence

View All
How Revolut Cut AI Coding Costs 90% With Self-Hosted Open Models - David Abitbol, Revolut

How Revolut Cut AI Coding Costs 90% With Self-Hosted Open Models - David Abitbol, Revolut

David Abitbol, who leads AI operations at Revolut, shares what AI tools did to 2,000 engineers' coding time and lead time. The main finding is that self-hosting open-source GLM models delivered about 60-70% faster delivery at roughly a tenth of frontier-lab costs. He covers the GPU sizing, latency targets, tuning that made the same hardware 88% faster, and how an EU-regulated bank keeps the setup compliant.

Grokking Software Architecture • Matt Erman & Luca Mezzalira • GOTO 2026

Grokking Software Architecture • Matt Erman & Luca Mezzalira • GOTO 2026

Matt Erman, author of Grokking Software Architecture, joins Luca Mezzalira to discuss why engineering fundamentals matter more in the AI era, how architects can act as clarity engineers, a five-step architectural thinking process, Architecture Decision Records, and the risks of letting AI generate systems without human judgment.

We Trust AI Agents That Break Every Day - Shy Ruparel, Temporal

We Trust AI Agents That Break Every Day - Shy Ruparel, Temporal

Shy Ruparel of Temporal breaks down the four gaps keeping AI agents from production, drawing on survey data from 554 engineers, and demos a crash-resilient $2,500 payout workflow built on durable execution with a local model.

Technology

View All
Mark Cuban’s Plan to Make Healthcare Affordable

Mark Cuban’s Plan to Make Healthcare Affordable

Mark Cuban and Julie Yoo discuss how transparency, direct contracting, and AI could reshape healthcare economics, from Cost Plus Drugs and modular pod manufacturing to personalized medicine, the future of physicians, and advice for founders who want to use AI to learn rather than to stop thinking.

Exploits Now Beat Patches: Open Source Security in the AI Era - Patrick Smyth, Chainguard

Exploits Now Beat Patches: Open Source Security in the AI Era - Patrick Smyth, Chainguard

Patrick Smyth of Chainguard examines how frontier AI has compressed the time between vulnerability disclosure and exploitation, leaving 30-day remediation cycles obsolete. The talk covers AI-assisted attacks, waves of supply chain malware, four attack surfaces spanning malware, known CVEs, zero days, and side doors like CI/CD, developer laptops, and agent skills, and practical defenses including cooldown periods, faster CVE remediation, and signing artifacts with SLSA and Sigstore.

Can Your AI Heat My House?

Can Your AI Heat My House?

A wide-ranging conversation on how AI inference is pushing cloud infrastructure, data centers, and hardware supply chains against physical limits, covering the rise of alt clouds, rack density and cooling, hardware lifecycles and circularity, talent shortages, heat reuse for communities, and what engineering leaders can do about resource-aware software.

Recent Post

Analyzing Group Chat Encryption in Messaging Applications

Analyzing Group Chat Encryption in Messaging Applications

This talk details a formal security analysis of group chat encryption algorithms in popular messaging applications like MLS, Session, and Keybase. It introduces Symmetric Sign Encryption (SSE) to model these protocols, identifying critical vulnerabilities such as insider replay and reordering attacks in MLS and Session due to insufficient context binding. The analysis highlights the complexities of key-dependent messages and key reuse, demonstrating how formal methods can pinpoint subtle design flaws and suggest robust mitigations for real-world secure communication.

Invited Research Talk: Measuring Generalization in EEG Foundation Models

Invited Research Talk: Measuring Generalization in EEG Foundation Models

This talk presents a multi-dimensional evaluation and interpretability framework for EEG foundation models. It reveals that current models often fail to outperform supervised baselines for BCI tasks, lack robustness to sparse channels, and exhibit an aperiodic low-frequency bias, making them better at capturing subject-specific rather than task-specific information. The analysis highlights critical deficiencies and suggests future directions for pre-training objectives and data collection to improve generalization.

TruthTable: A Verifiable Query Engine

TruthTable: A Verifiable Query Engine

TruthTable is a verifiable database engine that produces succinct cryptographic proofs for SQL query execution. It supports a wide range of SQL queries by leveraging query plans, polynomial encoding, and operator-specific PIOPs. It features a query planner with proof-specific optimizations and a novel batch compilation engine (ARCPOP). Benchmarks on TPC-H show average proving times of 55 seconds, verification times of 32 milliseconds, and proof sizes of 24kB, significantly outperforming prior academic and industrial systems in speed and expressiveness.

AI, Radio Astronomy, and the Search for Life Beyond Earth

AI, Radio Astronomy, and the Search for Life Beyond Earth

Ramiro Caisse Saide presents a multimodal deep-learning approach for technosignature detection in radio astronomy, using observations from Breakthrough Listen at MeerKAT. He investigates combining spectrograms with I/Q signal representations to improve signal detection and classification, particularly in low signal-to-noise environments. The talk also covers his extensive background in AI education, software development, the motivations behind SETI, fundamental radio astronomy concepts, and studies on Earth's own radio leakage.

AI models can now help run physical science experiments

AI models can now help run physical science experiments

The Model Hardware Standard (MHS) is a pioneering framework developed by Anthropic and HHMI Janelia to enable AI, specifically Claude, to safely and intelligently operate diverse scientific and manufacturing hardware. By abstracting device-specific communication, MHS dramatically accelerates scientific discovery, from automating complex microscopy tasks and real-time tracking to optimizing high-throughput drug screening, empowering researchers to focus on core scientific questions.

Can LLMs Write Fast Multi-GPU Kernels? — Simran Arora, Together AI

Can LLMs Write Fast Multi-GPU Kernels? — Simran Arora, Together AI

Simran Arora discusses the critical bottleneck shift in large AI workloads from GPU compute to inter-GPU communication. Her team's solution, ParallelKittens, offers a set of primitives to optimize multi-GPU kernels by leveraging fundamental transfer mechanisms and compute-communication overlapping. They introduce ParallelKernelBench, a benchmark to evaluate AI models' ability to generate such kernels, revealing that while models can handle syntax, they struggle with deeper reasoning about communication patterns and hardware trade-offs.

← PreviousNext →

Stay In The Loop! Subscribe to Our Newsletter.

Get updates straight to your inbox. No spam, just useful content.