From Software Engineer to AI Engineer – Janvi Kalra on The Pragmatic Engineer

Janvi Kalra discusses her transition from software engineering to AI engineering, offering practical advice for others considering the same path.

The Transition Journey

Background

• Started as traditional software engineer
• Became interested in AI/ML applications
• Made deliberate transition to AI engineering

Key Differences

• More experimentation required
• Less deterministic outcomes
• Faster-changing landscape
• Different success metrics

What AI Engineers Actually Do

Day-to-Day Work

• Building AI-powered features
• Prompt engineering and optimization
• RAG system development
• Evaluation and monitoring
• Integration with existing systems

Skills Required

• Software engineering fundamentals
• Understanding of ML concepts
• Prompt engineering
• System design for AI
• Evaluation methodology

The Learning Path

What Transfers

• Software engineering skills
• System design thinking
• Debugging mindset
• Production experience

What’s New

• ML/AI fundamentals
• Prompt engineering
• Evaluation techniques
• AI-specific patterns

Practical Advice for Transitioning

Getting Started

1. Build projects: Hands-on experience is crucial
2. Learn the basics: Understand how LLMs work
3. Follow the field: Stay current with developments
4. Network: Connect with AI engineers

Common Mistakes

• Overthinking the transition
• Waiting for perfect preparation
• Ignoring software engineering skills
• Focusing only on models

The AI Engineering Landscape

Current Opportunities

• Startups building AI products
• Enterprises adding AI features
• AI infrastructure companies
• Consulting and services

Skills in Demand

• RAG implementation
• Prompt engineering
• Evaluation design
• Production ML systems

Challenges in AI Engineering

Technical

• Evaluation is hard
• Non-deterministic behavior
• Rapid tool changes
• Integration complexity

Career

• Role definition varies
• Expectations unclear
• Measuring impact difficult
• Keeping skills current

Building AI Products

What Works

• Clear use cases
• Iterative development
• Strong evaluation
• User feedback loops

What Doesn’t

• Over-engineering
• Ignoring user needs
• Skipping evaluation
• Premature optimization

The Future of AI Engineering

Trends

• More specialized roles
• Better tooling
• Clearer career paths
• Mature best practices

Uncertainties

• How roles will evolve
• Which skills will matter
• Market dynamics
• Technology changes

Key Takeaways

1. Software engineering skills transfer - Don’t undervalue your background
2. Learning by doing - Build projects to learn
3. Evaluation is crucial - It’s the hardest and most valuable skill
4. Stay curious - The field changes rapidly
5. Community matters - Connect with others in the field