Salad
DevOpsOrchestrates AI workloads across global GPU pools.
Browse by the job,
not the category
Most people arrive knowing what they need done, not which category it lives in. 65 jobs grouped into 12 kinds of work. Pick the kind on the left, or search.
Tick Compare on any two to four cards to put them side by side.
Orchestrates AI workloads across global GPU pools.
Gives feedback on how LLM applications affect wellbeing, helping keep AI labs accountable.
A CRM that automates data entry and adds an AI co-pilot named Duke for daily tasks.
Route LLM traffic through one gateway, observe calls in traces. But lacks built-in prompt management.
AI experts in data science and machine learning.
Gooey.AI helps build multilingual AI solutions, but lacks integration with custom datasets.
Predicts win odds for each map in a FACEIT veto using a model trained on 30 million matches.
Upload and train custom computer vision models.
Soccersm AI analyzes game data to improve team performance. Simple but focused.
Tet Latvia offers a warning, Pathway helps protect your data but can be slow to set up.
Shape is your AI data analyst, querying databases in plain English.
Simplifies RAG and ML apps with Postgres and GPUs, but requires a VPC for more than one node.
Smarter Testing, Faster Deployments
Trace and evaluate LLMs with OpenTelemetry, manage prompts, and compare models.
Calculates win odds for FACEIT matches and can auto-veto the weakest map for you.
LM Studio lets you run LLMs offline on your laptop. Supports various models.
Expert in AI/ML solutions, Visage Technologies delivers efficient and compliant products with custom edge AI development.
Investigating AI trends for society.
Epoch AI offers insightful trends and statistics on AI advancements, but lacks real-time data updates.
Web-native LLMs in your browser.
Create PostgreSQL databases instantly.
LAION provides open datasets and models for machine learning research.
GovAI produces rigorous research and fosters talent for navigating AI governance.
Models, training data and machine learning.
Data scientists and ML engineers prepare data, train and evaluate models, run experiments and move working models into production. The tools here cover that pipeline: model hubs and libraries (HuggingFace), no-code and AutoML platforms (BigML, Lobe, Abacus.AI), GPU compute (Vast.ai, Lightning AI, Modal) and local LLM runners like LM Studio. Compare reproducibility and how easily you can export a trained model.
AutoML platforms such as BigML and Abacus.AI choose algorithms and tune settings for you, which is fast for standard tabular problems. Writing models yourself gives more control over features, architecture and evaluation. Many teams use AutoML for a baseline, then build custom models when it falls short.
Apps such as LM Studio download open models and run them on your own machine, so data stays local. You need enough memory for the model size you choose, and smaller quantised models trade some quality for speed. This suits private data and offline experiments.
Marketplaces and platforms such as Vast.ai, Lightning AI and Modal rent GPU time by usage instead of hardware ownership. Compare availability, startup time, storage and how they handle interrupted jobs. Keep checkpoints saved, since cheaper capacity can be taken away mid-run.