I’ve been working as a full stack developer with Laravel and the LAMP stack for a while now. About two months ago, I transitioned internally to work on some AI projects at my company.
Currently, my main tasks include:
Connecting LLMs using tools like LangChain and LangGraph
Utilizing LangSmith to monitor system performance
Implementing RAG systems with ChromaDB to minimize errors in AI responses
Developing with Django and FastAPI for the backend
I aim to further my knowledge in areas like:
Leveraging LangSmith for AI agents and tool integration
Fine-tuning AI models
Working with images and other forms of data
To be honest, it still feels like I’m mainly doing web development, but I’m incorporating AI APIs to enhance functionality. I hope to secure a real GenAI position within the next few months.
For those who are currently in GenAI roles, what does your average workday look like? Are you involved in the same activities I’ve mentioned, or is it a different experience altogether?
I’m quite new to this area and mostly driven by my interest. What topics should I concentrate on studying? Any valuable resources you could suggest?
You’re already doing GenAI work - you just don’t realize it. Everything you listed? That’s what I do at my company, minus the fancy title.
Here’s what nobody tells you: most GenAI jobs aren’t about training models from scratch. We build production systems that actually work. Your RAG implementation with ChromaDB is more valuable than you think.
What got me recognized was understanding data flow. Focus on how your embeddings get created, how retrieval quality affects responses, and why your RAG system sometimes gives weird answers. Learn to debug the entire pipeline, not just API calls.
Get comfortable with evaluation too. Set up A/B tests for different prompting strategies. Measure retrieval accuracy. Build dashboards that show when your AI’s being dumb.
RAG is foundational to most GenAI applications, so definitely worth understanding deeply:
Start thinking about cost optimization. Production GenAI systems burn through API credits fast. Companies love engineers who can make things work cheaper. Your web dev background actually helps here - you understand caching and efficient data handling.
You’re closer than you think. Just start calling yourself a GenAI engineer and own it.
You’re doing more specialized work than you think. RAG implementation plus LangGraph orchestration? That’s solid foundation stuff companies actually need. I broke into dedicated GenAI roles by diving deeper into model evaluation - BLEU scores, perplexity, custom evaluation frameworks. Here’s what I’d focus on: model deployment at scale (Triton or TorchServe) and retrieval evaluation techniques. LangSmith monitoring is valuable, but build custom evaluation pipelines too. Most GenAI roles want people who get both the integration layer AND the underlying model performance optimization. Don’t just implement RAG systems - become the person who can figure out why they’re not working.
You’re way closer to a real GenAI role than you think. RAG implementation, vector databases, LangSmith monitoring - that’s literally what most GenAI engineers do daily. Everyone thinks we’re training models from scratch or doing deep research. We’re not. I spend my days debugging retrieval systems, tweaking embedding strategies, and building APIs around LLM integrations. The difference between GenAI and regular web dev? Understanding how these systems actually work under the hood. Focus on data preprocessing for AI pipelines and learn embedding models beyond just plugging them in. Pick up evaluation metrics for generative systems too. Companies want people who can fix things when the AI breaks, not just when Django crashes. Your FastAPI/Django background is gold here - GenAI apps still need solid backends.
yeah, you’re def in a good spot! keep exploring those LLMs, and don’t forget about prompt engineering - it’s surprisingly key! also, learning about vector dbs can really up your game. wish you luck in your journey!
