Name: Mariya Mushtaq
Class Year: 2026
Major: Computer Science
Minor: Data Science
 
Internship Organization: Liberate Inc.
Internship Title:  Machine Learning Intern
Location:  Remote (USA)
 
What’s happening at your internship? We would love to hear what kind of work you are doing!  

This summer, I’m working as a Machine Learning Intern at Liberate, a fast-growing Series A startup building in the voice AI space. At this stage of my work, I’m working on curating high-quality training datasets for fine-tuning LLMs using real customer service transcripts – essentially turning raw, messy data into something that large language models can actually learn from. I’ve also been brainstorming logic to evaluate and structure this data in ways that directly impact how helpful and accurate the AI can be in real conversations. The project I’m working on is currently being experimented with to evaluate key metrics – and if successful, could help shape product decisions at a larger scale than I had expected.

Why did you apply for this internship?  

Liberate has been the kind of environment where things move fast, ideas are taken seriously, and there’s real space to learn by building. The team’s pace, ambition, and openness to experimentation have made it an ideal place to grow – not just technically, but also in confidence. There’s something energizing about knowing your work could make it into production and potentially shape how the product evolves. Being part of a small, focused team where everyone is working hard toward a shared vision has taught me a lot about autonomy, adaptability, and staying curious.
 
What has been your favorite part of this internship?   

Definitely the dataset work. I’ve really enjoyed the process of curating training data and developing logic to evaluate and structure it effectively. Thinking through how raw transcripts can be transformed into meaningful inputs for large language models – and how to design evals that actually reflect useful performance – has been both challenging and rewarding. It’s given me a much deeper appreciation for how critical high-quality data is in building reliable AI systems. Something that really stuck with me was a conversation where I mentioned my fear of breaking production. My manager laughed and said, “I broke production at my first job too – that’s how you learn,” and the CTO added, “If anything breaks, that’s on us, not you. Go ahead and explore.” That mindset has made it easier to take risks, experiment, and grow more confidently throughout the internship. 

What is something you have learned from your internship that you didn’t expect? 

One thing I didn’t expect to learn was just how important it is to be comfortable with being wrong. Coming into a fast-paced startup, I expected an intense, high-pressure environment where mistakes were to be avoided at all costs. I started this internship thinking I needed to get everything right on the first try, but I quickly learned that experimentation, failure, and iteration are not just accepted here; they’re expected. I learned that psychological safety isn’t just a buzzword; it’s a real strategy. Giving people the freedom to experiment and fail without fear is what allows a team to move quickly and build fearlessly. That realization was a significant mindset shift. It empowered me to take more initiative, try different approaches, and ultimately build more confidence in my decision-making. It’s been a powerful lesson in trusting the process of learning itself.