Skip to content

Applied ML And Domain Roles

Use this guide if you want hands-on model adaptation and domain specialization without going all the way into frontier-model research.

Included Roles

Role Layer Best Fit What Differentiates It
AI Data Engineer Layer 4 learners who like pipelines, datasets, and feature flow data movement and quality around AI systems
NLP Engineer Layer 4 text-heavy modeling and adaptation language-task depth beyond prompt-only systems
Computer Vision Engineer Layer 4 image and video understanding roles visual representations and multimodal design
Data Scientist (GenAI) Layer 4 experimentation and metric-driven iteration evaluation, analysis, and model/business trade-offs
Conversational AI Engineer Layer 6 assistants, voice systems, and multi-turn interaction dialogue state, context, and speech interfaces
AI Trainer / RLHF Annotator Layer 4 human-feedback and alignment workflows annotation quality, preference data, rubric discipline

Learning Path

Phase 1: Foundation

Complete Part 1 of the Learning Path first.

Phase 2: Shared Core

# Topic Note Priority Est. Time
1 Advanced fine-tuning advanced-fine-tuning Must 4h
2 LLM evaluation deep dive llm-evaluation-deep-dive Must 3h
3 Hallucination detection hallucination-detection Must 3h
4 Synthetic data and data engineering synthetic-data-and-data-engineering Must 3h
5 ML experiment tracking ml-experiment-tracking Must 2h
6 Data versioning for ML data-versioning-for-ml Must 2h

Phase 3: Role-Specific Emphasis

Role High-Leverage Notes Why
AI Data Engineer cloud-ml-services, distributed-systems-for-ai, llmops pipeline reliability and operational ownership
NLP Engineer rl-alignment, continual-learning, reasoning-models deeper model-adaptation and behavior work
Computer Vision Engineer multimodal-ai, computer-vision-fundamentals, diffusion-models visual understanding and generation depth
Data Scientist (GenAI) evaluation-and-benchmarks, classical-ml-for-genai, llm-landscape experiment design and decision support
Conversational AI Engineer conversational-ai, voice-ai, context-engineering multi-turn behavior and speech interactions
AI Trainer / RLHF Annotator rl-alignment, ethics-safety-alignment, evaluation-and-benchmarks feedback quality and alignment judgment

Phase 4: External Skills

# Skill Recommended Focus Priority
1 Dataset and labeling discipline schema design, annotation QA, leakage control Must
2 Experiment analysis confusion analysis, ablations, error slicing Must
3 Domain literacy product, support, healthcare, finance, or visual domain depending on target role Good

Skills Breakdown

Common Technical Skills

  • evaluation design and error analysis
  • model adaptation and data quality discipline
  • reproducibility across experiments and datasets

Differentiators By Role

  • NLP and conversational roles need stronger language-behavior depth
  • computer-vision roles need multimodal and image understanding fluency
  • data-oriented roles need pipeline and instrumentation reliability

Soft Skills

  • careful labeling and review habits
  • pattern recognition in failures
  • clear communication of uncertainty and experimental limits

Portfolio Project Ideas

Project Description Skills Demonstrated Difficulty
Domain adaptation benchmark compare prompt-only, RAG, and fine-tuned variants on one domain task evaluation, data discipline, adaptation trade-offs Medium
Multimodal assistant build a document or screenshot-aware assistant with measurable quality metrics CV basics, multimodal design, evaluation Medium

Interview Preparation

Review advanced-fine-tuning, llm-evaluation-deep-dive, computer-vision-fundamentals, and conversational-ai.

Common themes:

  • How do you know whether an improvement is real or noise?
  • When should you use fine-tuning versus retrieval, rules, or human review?
  • What failure patterns matter most in your chosen modality or domain?

Sources