Playbooks

Practical playbooks for the parts nobody teaches — what to learn, how to follow up after interviews, and how to thrive once you're in.

Metadata

Title: AI Tech Stack Choices for Career Growth
Last Updated: 2026-07-15T20:23:08.013673

Framework Choices

Llm Frameworks

Raw Apis

Cons

  • ›More code to write
  • ›Reinvent wheels
Note: OpenAI API, Anthropic API directly

Pros

  • ›Full control
  • ›No abstraction overhead
  • ›Always works
Career Impact: Good to know fundamentals
Recommendation: Understand fundamentals, use frameworks in practice

Langchain

Cons

  • ›Abstractions can be leaky
  • ›Moving fast (breaking changes)

Pros

  • ›Industry standard for LLM apps
  • ›Huge ecosystem
  • ›Most job postings mention it
Trend: Very hot
Market Share: 60% (2026)
Career Impact: ESSENTIAL for LLM roles
Recommendation: MUST LEARN for LLM engineers

Llamaindex

Cons

  • ›Smaller than LangChain
  • ›Fewer job postings

Pros

  • ›Better for RAG
  • ›Clean abstractions
  • ›Good documentation
Trend: Growing
Market Share: 30% (2026)
Career Impact: Important for RAG-heavy roles
Recommendation: Learn after LangChain

Deep Learning Framework

Jax

Cons

  • ›Smaller ecosystem
  • ›Fewer jobs require it
  • ›Steeper learning curve

Pros

  • ›Very fast
  • ›Functional programming style
  • ›Used by some research labs
Trend: Growing in research

Best For

  • ›Research roles
  • ›Google DeepMind
Market Share: 5% (2026)
Career Impact: Niche - only if targeting specific roles
Recommendation: Optional, learn if interested in research

Pytorch

Cons

  • ›Deployment can be tricky
  • ›Less mature production tooling than TensorFlow

Pros

  • ›Industry standard for research and production
  • ›Most job postings require it
  • ›Great community and resources
  • ›Pythonic and flexible
Trend: Growing

Best For

  • ›Research
  • ›Most ML engineering roles
  • ›LLM work
Market Share: 70% (2026)
Career Impact: Essential - learn this first
Recommendation: MUST LEARN

Tensorflow

Cons

  • ›Less intuitive than PyTorch
  • ›Losing mindshare
  • ›Fewer new projects

Pros

  • ›Still used at Google and some companies
  • ›Good production tooling (TF Serving)
  • ›TensorFlow Lite for mobile
Trend: Declining slightly

Best For

  • ›Google roles
  • ›Mobile ML
  • ›Legacy projects
Market Share: 25% (2026)
Career Impact: Nice to know, not essential
Recommendation: Learn after PyTorch if needed

Learning Strategy

Beginner Path

  • ›Month 1-2: Python + basics
  • ›Month 3-4: PyTorch + ML fundamentals
  • ›Month 5: SQL + Git
  • ›Month 6: LLM APIs + LangChain
  • ›Month 7-8: Build 3 projects using above
  • ›Month 9+: Add Tier 2 skills based on target roles

Avoid Shiny Object Syndrome

  • ›Don't learn every new framework that comes out
  • ›Master fundamentals deeply first
  • ›Learn new tools when you need them for a project
  • ›Frameworks change, fundamentals stay
  • ›YAGNI (You Aren't Gonna Need It) applies to skills too

Skill Prioritization

Tier 1 Must Have

Skills

  • ›Python
  • ›PyTorch (or TensorFlow)
  • ›Pandas, NumPy
  • ›SQL
  • ›Git
  • ›Jupyter notebooks
  • ›LLM APIs (OpenAI, Anthropic) for 2026
Description: Required for 80%+ of ML/AI jobs
Time Investment: 3-6 months to proficiency

Tier 4 Specialized

Skills

  • ›C++/Rust (for performance)
  • ›CUDA (for GPU programming)
  • ›TensorRT (for optimization)
  • ›Specific domain tools
Description: Only if role requires
Time Investment: Varies widely

Tier 3 Nice To Have

Skills

  • ›Kubernetes
  • ›Spark
  • ›Airflow
  • ›Ray
  • ›MLflow/W&B
  • ›React (for demos)
Description: Differentiators but not required
Time Investment: 1-3 months each

Tier 2 Highly Valuable

Skills

  • ›Docker
  • ›LangChain (for LLM roles)
  • ›Cloud (AWS/GCP/Azure basics)
  • ›FastAPI (for APIs)
  • ›Prompt engineering
  • ›RAG (for LLM roles)
Description: Opens many doors, worth learning
Time Investment: 2-4 months to proficiency