PUBLISHED AI RESEARCH AND PROJECTS

AI research and projects

The research, projects, and explanations behind the people. Read the source, inspect verified code and models, and find a collaborator.

Clear all filters

A source check confirms that a source supports the summary. It is not an independent replication or endorsement. Our methodology ↗

Showing 33–48 of 241 proofs of work · Kind: implementation · Newest first
Page 3 of 16
implementationChecked Sep 23, 2026

Hands-on Fine-Tuning Open Source LLMs for Production Tasks

Santiago Valdarrama · Underfitted

Created open-source walkthroughs showing practitioners how to prepare structured JSONL datasets, apply LoRA adapters via Hugging Face PEFT, and quantize models using bitsandbytes on a single GPU.

Scope & limitations

Single-GPU fine-tuning restricts effective batch sizes without gradient accumulation and activation checkpointing.

Context: Llama 3 8B, Mistral 7B, PyTorch, Hugging Face TRL and PEFT.

View mission
implementationChecked Sep 23, 2026

Inflection-1 & Pi: Empathetic High-Fidelity Conversational Agents

Mustafa Suleyman · Microsoft AI

Spearheaded the training and deployment of Inflection-1 and Inflection-2, powering the Pi conversational assistant with natural voice synthesis, low-latency conversational turn-taking, and empathetic persona tuning.

Scope & limitations

Empathetic conversational tone requires strict guardrails to prevent emotional over-dependence in vulnerable users.

Context: Proprietary foundation model trained on a 22,000 NVIDIA H100 GPU cluster with FP8 mixed precision.

View mission
implementationChecked Sep 23, 2026

SafeTensors: Fast, Secure Tensor Storage for Deep Learning

Julien Chaumond · Hugging Face

Spearheaded the creation and adoption of SafeTensors, replacing insecure Python pickle files with zero-copy memory-mapped deserialization that prevents arbitrary code execution and boosts model loading speeds 2-10x.

Scope & limitations

Requires strict contiguous memory layout incompatible with raw arbitrary python object serialization.

Context: Rust-based tensor serialization format adopted universally across Hugging Face, PyTorch, and vLLM.

View mission
implementationChecked Sep 23, 2026

Scale AI Data Engine for Supervised Fine-Tuning & RLHF at Scale

Alexandr Wang · Scale AI

Engineered human-in-the-loop expert annotation pipelines generating multi-million token pairs of complex instruction following, code review, and preference datasets powering the training of leading open and commercial models.

Scope & limitations

Human expert verification costs scale super-linearly with problem complexity in post-doctoral mathematics and legal reasoning.

Context: Distributed annotation platforms with automated inter-annotator agreement metrics and synthetic validation loops.

View mission
implementationChecked Sep 23, 2026

SEAL Leaderboards: Standardized Private Evals for Frontier LLM Benchmarking

Alexandr Wang · Scale AI

Created the SEAL evaluation platform providing contamination-free, expert-curated benchmark suites testing advanced programming (Coding-GSM), mathematical rigor, and adversarial safety to combat public leaderboard gaming.

Scope & limitations

Requires ongoing prompt refresh cycles to prevent indirect test data leakage into web-crawled pretraining corpora.

Context: Private held-out dynamic evaluation testbeds continuously updated by human domain experts.

View mission
implementationChecked Sep 23, 2026

Tenstorrent Wormhole & Blackhole RISC-V AI Accelerator Architecture

Jim Keller · Tenstorrent

Architected the Wormhole and Blackhole processor cards integrating Tensix array cores with native RISC-V scalar processors and on-chip Ethernet switches, enabling direct chip-to-chip 2D mesh networking for LLM tensor parallel training without external PCIe bridges.

Scope & limitations

Software compiler ecosystem (TT-Metalium) requires continuous software developer adoption to compete with NVIDIA CUDA.

Context: Scalable RISC-V and Tensix matrix cores running Llama 3 and DeepSeek MoE workloads over native Ethernet topology.

View mission
implementationChecked Sep 23, 2026

Tesla Full Self-Driving (FSD) Dual-Chip Computer Architecture

Jim Keller · Tenstorrent

Engineered Tesla's custom in-car FSD silicon featuring dual independent neural network accelerators processing 2,300 frames per second at under 72 watts, providing redundant lockstep execution for safety-critical real-time automotive vision.

Scope & limitations

Thermal dissipation in vehicle environments imposes strict wattage ceilings on sustained inference throughput.

Context: Custom dual NPU silicon operating on 8 camera streams at 36-teraops per chip.

View mission
implementationChecked Sep 23, 2026

The Covariant Brain: Universal AI for Industrial Robotic Manipulation

Pieter Abbeel · UC Berkeley / Covariant

Architected the commercial robotic perception and control system deployed across North American and European logistics warehouses, autonomously picking, placing, and bagging millions of previously unseen SKUs at 99.9% reliability.

Scope & limitations

Deformable objects (e.g. polybags, transparent shrink-wrap) require specialized tactile and multi-view sensor fusion.

Context: Multimodal vision-action transformer policies trained on real-world industrial sensor telemetry.

View mission
implementationChecked Sep 23, 2026

The Hugging Face Hub: Decentralized Open-Weight ML Infrastructure

Julien Chaumond · Hugging Face

Designed and scaled the open machine learning repository architecture supporting version-controlled model weights (SafeTensors), streaming dataset loaders, model cards, and automated evaluation leaderboards.

Scope & limitations

High bandwidth demands require aggressive regional edge caching and chunked tensor streaming.

Context: Global CDN serving petabytes of model checkpoints daily with SafeTensors memory-mapped loading.

View mission
implementationChecked Sep 23, 2026

The Machine Learning Reproducibility Checklist & Code Submission Mandates

Joelle Pineau · Meta FAIR / McGill University

Instituted the comprehensive standard adopted across NeurIPS, ICML, and ICLR requiring verified disclosures of random seeds, hyperparameter search spaces, computing infrastructure specs, and open code repositories to resolve the ML reproducibility crisis.

Scope & limitations

Checking hardware parity across multi-thousand GPU clusters remains difficult for smaller independent academic labs.

Context: Adopted across tens of thousands of peer-reviewed machine learning papers globally.

View mission
implementationChecked Sep 23, 2026

The Vesuvius Challenge: Machine Learning on Carbonized Papyrus

Nat Friedman · AI Grant / Andromeda Cluster

Co-founded and funded the global open scientific prize that mobilized computer vision and 3D deep learning teams to detect hidden ink patterns in non-invasively scanned 3D CT volumes of Herculaneum scrolls sealed in 79 AD.

Scope & limitations

Papyrus segmentation requires substantial manual inspection and geometric mesh flattening prior to virtual unrolling.

Context: 3D U-Nets and Vision Transformers processing gigavoxel synchrotron X-ray tomographic volumes.

View mission
implementationChecked Sep 23, 2026

TransformerLens: Open Source Mechanistic Interpretability Framework

Neel Nanda · Google DeepMind

Created and maintains the premier open-source library allowing researchers to hook into, intervene upon, and cache intermediate activations across 50+ autoregressive transformer architectures with PyTorch integration.

Scope & limitations

Activation patching on models larger than 70B parameters requires substantial distributed multi-GPU memory.

Context: Hook point engine supporting GPT-2, Llama, Pythia, Mistral, and custom transformer variants.

View mission
implementationChecked Sep 23, 2026

You.com Multimodal Conversational Search Engine & Real-Time Web RAG

Richard Socher · You.com

Engineered one of the first production conversational search engines combining real-time web crawling, zero-click answer synthesis, source URL grounding citations, and interactive code execution in a single chat interface.

Scope & limitations

Real-time web retrieval latency introduces latency overhead compared to static in-memory parametric generation.

Context: Hybrid BM25 and dense neural retrieval pipeline integrated with frontier LLMs and live web indexing.

View mission
implementationChecked Sep 23, 2026

AI Due Diligence Framework for Growth-Stage Corporate Acquisitions

Matt Higgins · RSE Ventures

Developed an institutional investment audit standard assessing target companies' defensibility against generative disruption, proprietary data moat validity, and internal AI workflow integration maturity.

Scope & limitations

Rapid foundation model capability leaps can render specific vertical software moats obsolete quickly.

Context: Private equity due diligence protocols, proprietary dataset evaluation, tech stack vulnerability audits.

View mission
implementationChecked Sep 23, 2026

AI Video Production & Voice Cloning Pipeline for Multilingual Media

Shelby Church · Shelby Church Media

Documented an automated media localization pipeline integrating ElevenLabs, HeyGen, and Descript to translate long-form technical video into 5 languages while preserving voice inflection and lip synchronization.

Scope & limitations

Complex idioms and regional humor still require human translator cultural inspection.

Context: Voice synthesis, automated transcript timing, AI generative video inpainting.

View mission
implementationChecked Sep 23, 2026

AI-Augmented Account-Based Marketing (ABM) Personalization Matrix

Nicole Leffer · Nicole Leffer Consulting

Built automated research and briefing workflows enabling enterprise sales development reps to tailor outreach to enterprise accounts using contextual AI data extraction from 10-K filings and executive interviews.

Scope & limitations

Public filing data can be up to 90 days lagging real-time corporate organizational shifts.

Context: SEC EDGAR API, PDF extraction, conversational synthesis, CRM integration.

View mission
Showing 33–48 of 241 proofs of work