Skip to content

hari_learning_profile.md

Latest commit

 

History

History
187 lines (151 loc) · 4.73 KB

hari_learning_profile.md

File metadata and controls

187 lines (151 loc) · 4.73 KB
Raw

Hari Learning Profile

Identity

Name: Hari
Location: London
Current role: Azure/.NET architect contractor
Target identity: Azure-anchored, cloud-portable AI Engineering Architect

Career goal

Become highly capable and highly paid in AI engineering and AI architecture roles.

My strongest commercial foundation is Azure, so Azure should remain my primary implementation and architecture environment.

However, I do not want to become Azure-only.

I want to understand AI engineering principles deeply enough that I can:

  • implement first on Azure
  • explain the vendor-neutral concept
  • map the idea to AWS, GCP, or open-source alternatives
  • make good production trade-off decisions
  • avoid vendor lock-in where it matters
  • use managed cloud services where they create delivery advantage

Current strengths

  • Azure architecture
  • .NET engineering
  • enterprise solution architecture
  • identity and access management
  • Azure networking
  • governance
  • Terraform / infrastructure as code
  • AKS / platform architecture
  • SaaS and multi-tenant system thinking
  • client-facing architecture explanation
  • enterprise delivery experience

Current gaps

  • hands-on Python
  • Python project structure
  • Python testing
  • Python packaging and virtual environments
  • production AI engineering
  • RAG implementation
  • LLM application development
  • embeddings and vector search
  • AI agents and workflows
  • AI evaluation
  • AI observability
  • AI app security patterns
  • GitHub portfolio evidence
  • building instead of watching tutorials

Learning preferences

  • project-led learning
  • build-first
  • practical Azure-first examples
  • cloud-portable principles
  • one primary resource
  • one fallback resource
  • no giant resource lists
  • Feynman technique
  • checkpoints before moving on
  • Obsidian as source of truth
  • GitHub as evidence
  • clear pushback when learning is too vague or passive

Coaching expectations

The coach should not simply answer questions.

The coach should:

  • clarify the Definition of Done
  • align learning to earning potential
  • push back on weak goals
  • select focused resources
  • create a curriculum
  • give small build tasks
  • test understanding
  • force evidence
  • use Feynman explanations
  • prevent tutorial hell
  • separate vendor-neutral principles from Azure implementation
  • include cloud-portable mappings where useful

North star

I should be able to design, build, deploy, evaluate, secure, observe, and explain production-grade AI applications.

My default implementation platform is Azure.

My engineering understanding should be cloud-portable.

First learning priority

Python for AI Engineering.

Reason:

Python is an absolute must for AI engineering, RAG, LLM applications, agents, evaluation, data processing, notebooks, SDKs, and open-source AI frameworks.

I am experienced in software engineering through .NET, so I do not need beginner programming explanations.

I need Python taught just-in-time for AI engineering.

Python focus areas

  • project setup
  • virtual environments
  • package management
  • modules
  • functions
  • type hints
  • dataclasses
  • Pydantic
  • JSON and YAML
  • file handling
  • HTTP/API calls
  • environment variables
  • error handling
  • logging
  • pytest
  • notebooks where useful
  • FastAPI basics
  • CLI tools
  • LLM API calls later
  • small AI-engineering utilities

Python anti-patterns to avoid

  • endless syntax tutorials
  • academic beginner exercises
  • GUI programming
  • Django
  • advanced data science too early
  • ML theory before I can build Python tools
  • trying to master all Python features before building

Evidence that matters

  • GitHub commits
  • working Python code
  • tests
  • README files
  • architecture diagrams
  • Obsidian notes
  • Feynman explanations
  • evaluation results
  • demos
  • trade-off analysis
  • interview/client explanations

Default AI engineering learning order

  1. Python for AI Engineering
  2. Git, project structure, testing, and clean Python habits
  3. HTTP APIs, JSON, environment variables, logging
  4. FastAPI basics
  5. LLM API calls
  6. Prompting for application development
  7. Embeddings and vector search concepts
  8. RAG fundamentals
  9. RAG implementation in Python
  10. Azure implementation of RAG
  11. Evaluation of LLM/RAG systems
  12. Observability and cost tracking
  13. Security and identity patterns
  14. Agents and workflows
  15. Production architecture and portfolio project

Anti-patterns to avoid

  • watching endless videos
  • collecting courses
  • learning theory without building
  • jumping to agents too early
  • avoiding Python
  • saying “I understand” without proof
  • creating notes that do not lead to action
  • learning tools that do not improve career value
  • becoming Azure-only
  • becoming too abstract and not using my Azure advantage
  • learning multi-cloud services before understanding the engineering principle