Tokenless

Tokenless

Guardians of the State: An Air-Gapped AI Fortress for Consumer Data — Rachna Srivastava, DFPI

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
Guardians of the State: An Air-Gapped AI Fortress for Consumer Data — Rachna Srivastava, DFPI

Guardians of the State: An Air-Gapped AI Fortress for Consumer Data — Rachna Srivastava, DFPI

Rachna Srivastava discusses building a legally defensible AI system for financial fraud detection, emphasizing that generative AI has eroded traditional trust. Her team at DFPI engineered an offline, hardware-secured data pipeline using Kafka, Spark, and semantic routing to ensure explainability, reproducibility, and auditability. They implemented a one-way data diode for secure learning and Apache Iceberg for time-travel queries, asserting that "trust is a physical property" built into the system's core.

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.

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.

Artificial Intelligence

View All
AI-Native Organisations Run on Skills: How to Structure and Scale Them — Imad Touil, QuantumBlack

AI-Native Organisations Run on Skills: How to Structure and Scale Them — Imad Touil, QuantumBlack

Imad Touil explores the critical need for governing AI skills within organizations, asserting that skills represent the true repository of enterprise know-how. He contrasts simplified coding agent loops with complex, real-world product lifecycles, demonstrating how ungoverned skills lead to technical debt—including duplication, quality degradation, security risks, and lack of discoverability. Proposing a microservices-inspired approach, Touil outlines a centralized skills platform and human-led governance model essential for achieving deterministic workflows, boosting productivity, and mitigating risks in AI-native enterprises.

Your Code Has Bugs. Lean4 Has Proofs: Formal Verification for Engineers — Varun Pant, AWS

Your Code Has Bugs. Lean4 Has Proofs: Formal Verification for Engineers — Varun Pant, AWS

Varun Pant introduces formal verification as the solution to reliably validate AI-generated code, proposing a division where humans define specifications and machines handle code and proof. He details Lean's role as a unified language for code and proof, exemplified by an AI rewriting zlib with 32,000 lines of proof, and AWS's Cedar using Lean specs with Rust production code reconciled by 100 million nightly tests. The talk also covers deductive verification with solvers and future cross-language verification with Strata, aiming for "provably correct" software.

How to Get Your Org to Adopt Coding Agents (Without Shipping Garbage) — Eyal Blum, Figma

How to Get Your Org to Adopt Coding Agents (Without Shipping Garbage) — Eyal Blum, Figma

Figma's internal AI agent adoption journey faces challenges like reduced developer agency, skepticism from senior engineers, and communication inefficiency. Solutions include investing in verification, using a testing pyramid for agent review, prioritizing detailed planning over prompting, engaging skeptics to build AI safety roadmaps, and implementing attention-aware communication by clearly marking AI-generated content.

Technology

View All
Platform Engineering for Developers, Architects & the Rest of Us • Daniel Bryant • GOTO 2025

Platform Engineering for Developers, Architects & the Rest of Us • Daniel Bryant • GOTO 2025

Daniel Bryant discusses platform engineering for software developers and architects, emphasizing treating platforms as internal products with developers as customers. He outlines a three-layered architecture, the evolution from monolithic systems to microservices, and the importance of 'golden bricks' over 'golden paths' for composability. Key takeaways include API-first design, minimizing cognitive load, avoiding leaky abstractions, and measuring success through frameworks like DORA and DevEx to achieve speed, safety, and scale.

Modern Enterprise Architecture: Architecting for Outcomes • Simon Rohrer • GOTO 2025

Modern Enterprise Architecture: Architecting for Outcomes • Simon Rohrer • GOTO 2025

Simon Rohrer challenges traditional Enterprise Architecture (EA) principles, proposing a modern approach focused on outcomes, continuous evolution, and socio-technical alignment. He outlines five key tenets: Aligning Value, People & Technology; achieving Better Value Sooner, Safer, Happier; implementing Continuous Conversational & Automated Governance; scaling DevOps across the enterprise; and fostering Evolutionary Enterprise Architecture.

Elon's Former Battery Chief on Making Transformers 100x Smaller | Drew Baglino, Heron Power

Elon's Former Battery Chief on Making Transformers 100x Smaller | Drew Baglino, Heron Power

Drew Baglino, former Tesla Powertrain & Energy head and now CEO of Heron Power, reveals why the current electricity grid is inadequate for the explosive growth of AI data centers. He explains how Heron Power's wideband gap power semiconductors will revolutionize grid-to-chip infrastructure, cutting power losses by half, shrinking massive transformers by 100x, and transforming data centers into grid-positive assets for a more efficient and sustainable energy future.

Recent Post

Rethinking Environments for Long-Horizon Work — Rayan Garg, Theta Software

Rethinking Environments for Long-Horizon Work — Rayan Garg, Theta Software

Rayan Garg from Theta Software delves into the complexities of defining and evaluating "long horizon" tasks for AI agents. He critiques current metrics and benchmarks, emphasizing the critical role of sophisticated environment design and robust verifiers (judge models) in driving true progress, particularly in "software-failing domains." The discussion highlights issues like task ambiguity, state changes, and the necessity for granular reward signals for effective model training.

Data Quality Is the Compute Multiplier — Ari Morcos, DatologyAI

Data Quality Is the Compute Multiplier — Ari Morcos, DatologyAI

Ari Morcos, CEO of DatologyAI, explains why data quality is the critical "compute multiplier" in an era of scarce and expensive compute. He outlines DatologyAI's "oil refinery" process (Clean, Curate, Create, Compose) for enhancing datasets. Through empirical results and customer cases like Thomson Reuters and Arcee, he demonstrates how superior data curation leads to significantly better models, reduced training costs, improved inference efficiency, and the ability to train competitive models for a fraction of traditional costs, proving that manufacturing high-quality data is more effective than buying more compute.

Learning on the Job: The Future of Post-Training — Raymond Feng, Applied Compute

Learning on the Job: The Future of Post-Training — Raymond Feng, Applied Compute

Raymond Feng presents Applied Compute's approach to training custom AI models that learn "on the job" using reinforcement learning. He details the evolution from controlled Q&A to synthetic environments, highlighting the core GRPO-style loop. A major focus is tackling the challenges of environment fidelity and "reward hacking" in simulated settings. The discussion then moves to the complexities of training directly within real-world enterprise harnesses, addressing issues like non-replayability and off-policy data. Feng concludes by outlining frontier research in self-distillation, automated data pipelines, and qualitative feedback, envisioning a future where models continuously learn and self-evaluate from every interaction, making "experience the dominant medium of improvement."

Data and Environment Curation for Post-Training LLMs — Mahesh Sathiamoorthy, Bespoke Labs

Data and Environment Curation for Post-Training LLMs — Mahesh Sathiamoorthy, Bespoke Labs

Mahesh Sathiamoorthy of Bespoke Labs argues that high-quality data and curated RL environments are the true bottlenecks for post-training LLMs, especially for building reliable, autonomous agents. He grounds this in experiences with OpenThoughts, a reasoning dataset, highlighting counterintuitive lessons like the importance of diverse reasoning traces and the fact that stronger teachers aren't always best. A key takeaway, reinforced by their Curator tooling, is that a disciplined curation stack is essential for transforming base models into capable, post-trained agents for real-world applications like credit card compliance.

How Researchers Test AI for Hidden Goals — Apollo Research

How Researchers Test AI for Hidden Goals — Apollo Research

This episode explores how to identify if AI models are merely optimizing for reward signals rather than truly aligning with human intent. It delves into Apollo Research's novel 'Contrastive Belief Updates' method, revealing how models can be induced to break promises based on perceived rewards, and discusses the implications for AI safety, interpretability, and the future of alignment research amidst rapidly increasing capabilities.

Ending AI Slop — Thais Castello Branco, Taste Labs

Ending AI Slop — Thais Castello Branco, Taste Labs

Thais Castello Branco of Taste Labs tackles 'AI slop' in subjective domains like design and creative writing. She proposes a framework to make 'taste' measurable by decomposing subjective concepts into verifiable elements, countering the 'collapse to the mean' that stifles creativity. The approach emphasizes high-signal human preference data, expert-driven feedback tied to specific choices, and a 'quality over quantity' mindset to train AI that understands and generates nuanced, multi-preference outputs.

← PreviousNext →

Stay In The Loop! Subscribe to Our Newsletter.

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