Hiro Agentic Security - Hiro is your agentic security team — scan, fix, triage, and turn every fix into SOC 2 evidence. Built for the Claude Code stack, it works from first commit to audit. Hiro is the autonomous security team for fast-moving companies, built by former SRE leaders at Samsara who spent years doing this work manually. Y Combinator-backed.

Thinnest AI - ThinnestAI is BYOK programmable voice AI infrastructure for building human-like AI phone agents at scale. Built in India for global teams, it supports 100+ languages (every official Indian language plus global) with native STT + TTS pipelines per language — not English-with-translation. Bring your own LLM, STT, TTS, and telephony providers. No-code flow builder with RAG knowledge bases and MCP support. Flat ₹1.5/min platform fee with INR billing, native SIP and Twilio integration, REST + Python + JS SDKs. Targeted at Indian banks, NBFCs, insurers, BPOs, edtech, healthtech, D2C, and logistics companies.

Trainer - Trainer lets you train AI agents by simply recording a task once — no prompts, no labeled data. It captures every click, keystroke, and intent as you work, then turns that workflow into a reusable agent that can repeat the process reliably. Trainer doesn't rely on fixed coordinates or pixel positions; it builds a semantic understanding of the UI (what the element is, surrounding context, labels, and the intent behind the action) so the agent can still find the right element when buttons move or layouts shift. Works at the level of intent, not rigid steps, so it can adapt when small variations appear in the flow.

Multi-Claude - Multi-Claude lets you run multiple Claude Desktop accounts side by side on one Mac. Each account is a real dock app with its own icon, history, and login. It detects your existing Claude install and offers to import your history into a named profile, keeping the original as backup. Profiles live entirely on your Mac — no accounts to create, no data sent anywhere. Name each profile (Work, Personal, Side project), pick a color, and switch between them with Cmd-Tab like any other app. Multi-Claude is a thin wrapper that launches the real Claude app with a different data directory.

Owlish - Owlish turns your knowledge base into AI customer support agents that answer questions, take support actions, and hand off to your team across web, Slack, Teams, and more. Train it on your website, help center, PDFs, DOCX, Markdown, or resolved tickets — Owlish only answers from the sources you've ingested and every reply cites the exact passages it drew from. If the answer isn't in your knowledge base, it says so and offers a handoff instead of guessing. Most teams go from zero to a live, grounded agent in under ten minutes.

Software Engineer, Secrets Infrastructure (Remote) - Fully remote team building Stripe's internal secrets/credentials infrastructure.

Software Engineer, New Grad - New grad SWE role spanning Stripe's payments, money-movement and platform teams.

Applied Scientist / Research Engineer, AI4Engineering - Apply Mistral's models to coding, software engineering, and developer workflows.

AI Scientist - Research scientist role on Mistral AI's research org — onsite, hybrid or remote.

Research Engineer, Machine Learning - Build and optimize large-scale learning systems powering Mistral's open-weight models.

Machine Learning Engineer, Fast Optimized Inference (EMEA Remote) - Build the fastest open-source inference stack — quantization, kernels, serving — at Hugging Face.

Cloud Machine Learning Engineer (US Remote) - Build Hugging Face's cloud ML offerings — inference endpoints, training infrastructure, and customer-facing tooling.

SDE AI/ML II — ML Inference Apps, AWS Neuron - Performance tune and enable ML model families (Llama2, GPT-class) on AWS Neuron / Trainium inference accelerators.

Annapurna Labs at AWS — Early Career (US): ML Systems & Silicon Innovation - Multiple technical tracks across ML Systems & Compilers, Distributed Training Systems and ML Infrastructure on AWS's silicon team.

Machine Learning Research Engineer, Generative AI (Siri) - Build infrastructure, datasets and models that empower Siri across NLU, dialog generation, speech synthesis and multi-modal interaction.

AI Research Scientist — Safety Alignment Team - Research and develop safety-alignment techniques for Meta's frontier AI systems.

AI Research Scientist, Language — Generative AI - Conduct research on generative language models within Meta's GenAI org.

Senior Deep Learning Algorithm Engineer - Develop and optimize deep learning algorithms for NVIDIA's accelerated computing platforms.

Deep Learning Software Engineer, TensorRT Performance — New College Grad 2026 - Integrate TensorRT into open-source frameworks and implement graph compiler algorithms across NVIDIA's inference ecosystem.

Research Engineer, Pretraining - Build and optimize pretraining systems and pipelines for frontier models at Google DeepMind.

Research Engineer, World Models - Implement core infrastructure and conduct research to build generative models of the physical world.

Staff Machine Learning Engineer, Virtual Collaborator - Lead ML engineering for the Virtual Collaborator product line at Anthropic.

Research Engineer, Cybersecurity Reinforcement Learning - Safely advance model capabilities in secure coding, vulnerability remediation, and defensive cybersecurity.

Research Engineer / Research Scientist, Pre-training - Develop the next generation of large language models on Anthropic's Pre-training team.

Research Engineer / Scientist, Alignment Science - Contribute to exploratory experimental research on AI safety with a focus on risks from powerful future systems.