I’ve officially applied to 100 jobs as a new grad! It’s fair to say this workflow has been reduced to an atomic science. As a computer scientist and software engineer, I’ve tried to reduce as much friction in this process as possible by abstracting and automating away tasks that cause me the most anxiety and dread. Now, that isn’t as simple as just asking AI to come up with a response for everything I come up against, although that would make my job search a lot easier.
In the process of trying to automate every process in my workflow, from search to application, I kept facing this urge to conform and flatten my profile for the sake of application convenience. If every application I send is the same, it makes it easier to automate. The issue is, I have to be willing to sacrifice my unique and once-in-a-lifetime profile to cut corners and save time. That’s something I just can’t live with. Read more about my job search process here
One habit I’ve created to save time on applications without sacrificing my own voice is to keep a running document of all of the personal application questions I’ve answered in my applications. I plan to converge on a handful of responses that have been curated for my applications. It’s also practice for my behavioral interviews to meditate on what kind of candidate I want to be seen as.
I realized recently that this could actually be a good portfolio piece for my recruiters as well; why wait for the job application when you can find my responses here? So, without further ado, here are my most frequently asked questions! If I’m going to be reduced to one personality, let it be this one. These are the questions that I found to be most insightful to who I am as a worker, and reveal some details about my work that I like to lead with. Feel free to reach out if your company is hiring; I’d love to discuss how I could fit into your staff.
What are you looking for in your next role? What would you like to avoid?
What I’m looking for most is mentorship. I’ve proven to myself that I can learn independently, taking projects from class notebooks to deployed apps on my own, but I know the fastest way to grow from a good engineer into a great one is working alongside people who’ve already made the mistakes I haven’t yet. I want a team where code review is teaching, not just gatekeeping, and where I can eventually pass that mentorship forward. What I’d like to avoid is an environment that treats burnout as a badge of honor. I work intensely and take real pride in my output, but I’ve learned that sustainable pace and honest feedback loops are what keep that intensity productive over years rather than months.
How intensely do you like working?
I don’t like working against the tide; I streamline my workflows to prevent friction and in pursuit of what I call “flow state.” I find joy in everything I do, so my work never feels like a chore. When I’m in flow state, my work practically does itself, and I get lost in my tasks, however boring. I’m a bit of a perfectionist, so I’m willing to run the extra mile if it means the final product meets my own standards.
What do you optimize for in life?
I optimize for community. I have a really bad case of FOMO (fear of missing out), so I constantly bend over backwards to ensure I can be present for every major social event my friends put on. I feel most at home in a crowd, and I love meeting new people. When I’m not making time for friends, I’m in my local community working on mutual aid projects.
If we gave you a vague problem, how would you figure out what to build?
I got a taste of this problem during my study abroad in Ghana: the nonprofit we were working with couldn’t tell us what to build; they just knew their digital tools weren’t working. By digging into how their site and tools were actually being used, we found the real problem was one nobody had named, which was that their staff couldn’t update their own website. That became the product. So my process would be: define what solved looks like, gather evidence from the people experiencing the confusion before building anything, narrow it into two or three testable hypotheses, and ship the smallest thing that validates the strongest one.
Tell us about a product or project you built end-to-end
I recently rebuilt my portfolio site from a static Create React App into an Astro.js blog, owning everything from design to deployment. Beyond just showcasing projects, I built it as a writing platform for case studies, build logs, and personal essays, because I believe documenting the process is as valuable as the product. The rebuild meant migrating content, redesigning the information architecture, and learning Astro’s content collections along the way. It’s the project that best represents how I work: iterative, documented, and built to grow.
Share a project you’re especially proud of and explain what made your contribution meaningful
I led a 5-person team building a wardrobe management app with AI-powered outfit recommendations tailored to weather and personal style. My contribution was the data ingestion pipeline: scraping, cleaning, and embedding catalog data from Macy’s and Depop. It was the backbone of the product, since the recommendation feature couldn’t exist without it. I’m proud of it because it forced me to own an entire subsystem end-to-end and hand off clean, usable data to my teammates.
Please share 1-3 links to any relevant work
- My Portfolio doubles as both my portfolio and personal blog. I’ve grown to enjoy documenting my creative process, so this will become the home for those writings.
- Tennis Oracle started as a Jupyter Notebook from a class project that set out to predict the winner of ATP Tennis matches; I since revamped it into a deployed web app that predicts current ATP Tennis matches.
- Virgo is an agentic wardrobe management app that helps users curate outfits, schedule outfits to their calendar, and find gaps in their wardrobe, all using a familiar chat interface. Built using Google ADK and MongoDB Atlas Vector Search.
Please describe your AI experience
My AI experience spans formal coursework, building with models, and building alongside them. I studied at the University of Florida during its push to become “the AI university,” backed by a major partnership with NVIDIA and its co-founder Chris Malachowsky that put the HiPerGator supercomputer and AI coursework at the center of campus. As part of that, I completed UF’s AI Fundamentals Certificate, which included a graduate-level machine learning fundamentals course, easily the most challenging and rewarding coursework of my degree. On the applied side, I built Tennis Oracle, a match prediction model for ATP tennis that started as a class Jupyter Notebook and is now a deployed web app predicting current matches. More recently, I built Virgo, an agentic wardrobe management app using Google’s Agent Development Kit and MongoDB Atlas Vector Search, where users curate outfits, schedule them to their calendar, and find wardrobe gaps through a chat interface. That project taught me the practical realities of agentic systems: tool design, retrieval quality, and keeping an agent grounded in real user data. Finally, AI is central to how I work day to day; Claude writes most of my code while I focus on problem definition, architecture, and review.
Why do you want to work at ______?
Because I like money, I like being paid, and I like being able to pay for my housing and to feed myself. I know that obtaining employment at your company would earn me quality benefits and a salary to pay off my student loans, so I would like to work at your company as soon as possible. I’m not desperate, but please hire me.
What are some AI-specific technologies you are comfortable with?
I’m comfortable across a few layers of the AI stack. For agentic and LLM applications: Google’s Agent Development Kit running Gemini, which powers my wardrobe management app Virgo, plus the OpenAI API for chat and embeddings. For retrieval: MongoDB Atlas Vector Search and OpenAI embeddings, which I’ve used to power semantic search over product catalogs and wardrobe data. For classical machine learning: scikit-learn and pandas, which I used to build and train Tennis Oracle, my ATP match prediction model, originally prototyped in Jupyter Notebooks. I’m also fluent with AI-assisted development tools, with Claude at the center of my day-to-day coding workflow.
How have you used AI tools while building a project?
Claude is at the center of my workflow. It writes most of my code while I handle problem definition, context management, architecture decisions, and review. The clearest example is Tennis Oracle: I used this workflow to take a class Jupyter Notebook and turn it into a deployed web app predicting live ATP matches. What worked well was treating the model as a collaborator with a spec rather than a magic box, with clear intent, tight iterations, and reviewing everything before it ships. I can direct and audit confidently because I’ve written plenty of code by hand and understand what the model is producing. What I’ve learned doesn’t work is vague prompting and blind acceptance; the output quality tracks the quality of the plan behind it.
What signs would tell you that AI-generated code needs improvement, even if it seems correct at first glance?
I look for the failure modes AI code is prone to: hallucination, over-engineering, inconsistent reasoning, and missing edge cases. As always, it takes a real user to know how an application is failing. Correct-looking is the starting point of review, not the end of it. The more AI code is written into a code base, the better code coverage that code base requires.