Job paths, salary insights, interview prep, projects, and portfolio templates — everything you need.
Salary ranges based on US market data, 2025. Remote rates similar.
Build LLM-powered products. Work on RAG, agents, prompt engineering, and production AI systems.
Train models, optimize inference, deploy ML systems at scale. Strong systems and infrastructure work.
Analyze data, build models, drive business decisions through experimentation and statistics.
Push the frontier. Publish papers, design new architectures, contribute to foundation models.
Specialize in prompt design, evaluation, and LLM behavior. Newer role — high variance in scope.
Define AI features, work with engineers and researchers, ship AI products that users love.
Design end-to-end AI systems for enterprise clients. Mix of consulting and engineering.
Build the data pipelines that feed AI systems. ETL, feature stores, vector databases at scale.
YouTube, courses, newsletters. Highly variable — top creators earn six figures from ads + sponsorships.
Standard LeetCode-style problems (medium difficulty), plus AI-specific: implement attention from scratch, vectorize a search, build a tokenizer. Practice 100+ problems before applying.
Backprop, regularization, transformer math. Be ready to whiteboard.
Design a recommendation system, RAG pipeline, or training infrastructure.
STAR format. Have 5 stories ready: leadership, conflict, failure, impact, growth.
Walk through your projects. Be ready to defend every design decision.
Recruiters skim portfolios in 30 seconds. These 8 projects signal different skills.
RAG over your notes/email/calendar. Shows you can build end-to-end LLM apps.
Upload a PDF, ask questions. Classic RAG starter — make it production quality.
Stable Diffusion + ControlNet. Fine-tune a LoRA on your own dataset.
Whisper + LLM + TTS. Phone agent or meeting bot. Shows multimodal skills.
Build a transformer (or CNN, or diffusion model) from scratch in PyTorch. No frameworks.
Pick a paper, reproduce results, write blog post. Strong signal for research roles.
Multi-agent system using LangGraph or CrewAI. Solve a real workflow problem.
Deploy via FastAPI + Docker. Add monitoring, evals, cost tracking. Show ops chops.
1. Top — Headline: Your role + 1-line value prop. 2. Projects section goes ABOVE work experience for new entrants. 3. Quantify everything: "Reduced inference latency by 40%" not "improved performance." 4. Skills: List concrete tools, not buzzwords. 5. One page max unless you have 10+ years.
"Familiar with Python" · "Worked with team" · No GitHub link · 3-page resumes · No metrics · Generic objective statements.
GitHub link prominently. Live demo URL. Specific tools (LangChain, vLLM, Triton). Measurable impact. Recency matters.
aijobs.net · WellFound · Y Combinator's Work at a Startup · Hugging Face Jobs
Google, Meta, Microsoft, Amazon, Apple — apply 6 months before you want to start.
OpenAI, Anthropic, DeepMind, Mistral, xAI, Cohere — competitive but worth trying.
LinkedIn (set alerts), Indeed, Otta. Apply within 24 hours of posting.
50% of jobs come through warm intros. Build in public. Engage on Twitter/LinkedIn.
Remote OK · We Work Remotely · Remote Year. Many AI startups are remote-first.