When I first started applying for AI and algorithm jobs—especially those related to large language models—I didn’t realize how much impact a resume could have. I thought having a solid background and showing “passion for AI” would be enough. So I put together a resume packed with technical terms, a long list of tools I’d used, and some high-level project descriptions. But honestly, it didn’t lead to many interviews.
Eventually, after speaking with a few friends who had received offers from reputable companies, I realized I was approaching it the wrong way. The first big thing I changed was how I described my projects. Take one I worked on—a document-based Q&A system using RAG. Initially, I wrote something like, “Built a knowledge Q&A system with LangChain and LLMs.” But looking back, that was way too vague. I rewrote it to explain what I actually did: how I used FAISS to create the vector store, how I designed prompts to handle multi-turn questions more effectively, and what metrics improved after deployment (like latency, recall, and stability). That made a big difference.
Another shift was moving away from buzzwords and focusing more on results. It’s easy to say you’re “familiar with Transformers” or “have experience with LoRA,” but unless you explain how you used them and what the outcome was, it doesn’t mean much. So instead, I started writing things like “Used QLoRA to fine-tune a model on custom data, improving BLEU by X,” or “Applied 4-bit quantization to reduce inference time by Y%.” These details feel a lot more grounded and tell the reader that you’ve built something.
One thing I underestimated early on was how much companies care about engineering ability. Even if the role says “ML” or “AI,” many teams want people who can ship code. That’s why I began highlighting things like “Deployed model as a service using Docker and AWS,” or “Optimized inference with vLLM to support multiple concurrent users.” It shows that you can do more than just run experiments locally.
Lastly, I worked hard to make the writing sound like me. Not overly polished, but clear and honest. I stayed away from generic phrases and focused on describing what I did in a way I’d explain to another engineer. After each version, I’d send it to a few friends to see what made sense and what didn’t. That process helped a lot.
If I had to sum it up, listing every tool or framework you’ve touched won’t get you far. What matters is whether you can show what you’ve built, how you solved real problems, and whether the results were meaningful. Adding a bit of engineering perspective and keeping the writing real can make your resume stand out.
Release time:2025-06-12
