Practicum Spotlight
The MSDS program at USF has a strong emphasis on hands-on learning. One of the unique aspects of the program is the chance to engage in a part-time internship lasting nine months, commonly referred to as a practicum. I had the pleasure of meeting with some students to learn about their practicum experiences. We’ll dive into some highlights in this article.
Rishi Mohan, Machine Learning Engineer at Metaphor Data
Q: “What projects have you been working on so far?”
A: “I’ve been working on building connectors for external knowledge-bases like Notion and Sharepoint, along with the backend for handling these external search documents. These document crawlers retrieve documents (like FAQs, knowledge spikes, design docs, etc.) from these sites. Thus, when a user makes a query with Metaphor AI, we can produce a better-contextualized and more accurate response using resources across their documents and integrate information at a deeper level.”
Q: “What are some new technical skills that you have learned?”
A: “I’ve definitely become much more familiar and comfortable with the development cycle and CI/CD (continuous integration and continuous delivery), and both of these have pushed me to write high quality code. I’ve also learned some TypeScript and worked more with LlamaIndex and LlamaHub loaders for reading and wrapping data for LLMs, especially focused on generating and using vector embeddings for similarity search.”
Q: “Has your practicum experience changed your career outlook”
A: “Working with the Metaphor team has made me more open to exploring roles in Data Engineering (and taking the Data Engineering extension!) since I much more appreciate the work that goes into writing production-ready applications and pipelines.”
Rashmi Panse, Machine Learning Engineer at How We Feel
Q: “Can you describe the projects you worked on so far?
A: “The data pipeline project is one I took lead on as I noticed a need for better accessibility to event log data that was being stored in MixPanel, and was quite tedious and computationally expensive to pull. Using the MixPanel integrations I was able to create various pipelines which map MixPanel data, stored as JSON, to a BigQuery database of tables. The pipeline is expected to save 100+ engineering hours yearly!
Currently, my practicum partner Jay and I are working on a two-step project with the aim of improving the personalized experience on the How We Feel app by leveraging user activity data. The first step involves clustering the user activity data to form cohesive groups of individuals with similar activity patterns on the app. This will allow us to have a better understanding of the segments of users that exist on the app. The second step is building a recommender system that leverages the clusters created to make better recommendations of the lesson videos and tools that exist in the app.”
Q: “Has your practicum experience changed your career plan/outlook?”
A: “This practicum experience has further solidified my interest in Machine Learning Engineering/Data Science. Prior to this role I wasn’t sure how to define the differences between a Data Engineer, Machine Learning Engineer, and Data Scientist. However through this practicum experience and coursework I feel like I can confidently say I’m interested in MLE/DS because of its combination of statistics and algorithm focus and innovative opportunities. I love how end-to-end my experience has been.
This role has definitely opened me up to start-up life more. I have never worked at a startup before, prior to this I was quite averse to joining a start-up, however the small team structure has allowed me to work very closely with extremely skilled and talented team members like the How We Feel Founder Ben Silbermann (also a Co-Founder of Pinterest), which has a great learning opportunity.”
Q: “Do you work in-office or remote?”
A: “My teammate and I mainly work remotely but we occasionally have in-person work sessions with our mentor. Recently we also had an on-site day with the full How We Feel team to set our vision and plan for 2024. It was such a great experience meeting everyone in person and also interact socially over a team dinner!”
Inseong Han, Data Scientist at LexisNexis
Q: “How has your experience been working in the office?”
A: “Working at the office has been a great experience. I have a dedicated workstation equipped with my own monitor, fostering a focused work environment. The office facilities include multiple meeting rooms and a private room for more secluded work. Additionally, amenities like a stocked refrigerator with coffee and soda, as well as a snack bar, provide convenience for daily needs. The office community is welcoming, with each employee demonstrating kindness and approachability, making it easy for me to seek assistance or ask questions.”
Q: “Who is your company mentor?”
A: “I am mentored by Yuhan (Hanna) Wang, who serves as a Lead Data Scientist. Throughout my onboarding, she offered invaluable support by guiding me through essential tasks, including acquiring access/permissions for company tools and gaining a comprehensive understanding of the code base used in the data pipeline. Fun fact, Hanna is also a USF MSDS alum. Her understanding of how hectic and challenging the USF MSDS curriculum is has been emotionally supportive, making me more comfortable to ask questions, especially as a student.”
Q: ”Has your practicum experience changed your career plan/outlook?”
A: “Indeed, my practicum experience has significantly influenced my career plan and outlook. Prior to the practicum, my professional background was centered around Computer Vision and Recommendation Systems.
My wife is pursuing her PhD in Linguistics at Stanford, and this exposure ignited my curiosity about NLP, ultimately leading me to explore NLP for my practicum. This choice allowed me to explore the intricate world of NLP applications and contemplate how real-world challenges could be effectively addressed using NLP techniques. The experience at LexisNexis became a pivotal moment in solidifying my decision.
Throughout the project, I observed a clear need for NLP techniques to navigate vast amounts of legal data efficiently. It became evident that leveraging LLM, such as ChatGPT, could provide practical solutions to challenges presented by terabytes of information. This realization underscored the power of NLP in a world dominated by textual data.
The practicum experience has been immensely enjoyable, prompting me to seriously consider a career path in NLP, where I can continue to explore creative applications and contribute meaningfully to addressing complex challenges”
Bassim Eledath, Machine Learning Engineer at Square
Q: Who is your company mentor? How often do you connect and what kind of help do they provide?”
A: “Heli Gong is my company mentor. She is a Senior Machine Learning Engineer at Block, based in Canada. We connect on twice a week to keep track of project work. Her guidance is primarily domain-related, focusing on understanding the tables and source data better to solve the problem I’m working on. When it comes to the methods used to solve the problem, I have a lot of autonomy in terms of choice and implementation.”
Q: “Can you describe a project you worked on so far?”
A: “The problem I’m currently working on, and nearly finished with, involves developing a ranking algorithm to determine which prospective merchants should receive direct mail as opposed to the less expensive email. This presents a challenging low signal-to-noise problem, further complicated by class imbalance due to most merchants not converting. I came up with a simple yet effective solution that has proven to increase Net Profit per merchant by around 60% compared to the current model in production. I am now in the process of implementing it in production and potentially A/B testing it.”
Q: “What is your career plan/outlook post graduation?”
A: “I definitely want to continue working in a role that offers a good mix of engineering, research, and implementation. On the other hand, I am not tied to a particular domain or even a subset of machine learning methods. Gen AI has opened a floodgate of new job opportunities, and I’ve been tinkering a lot with LLM and RAG applications, so I’m exploring those possibilities as well.”
