Myles Phung Email & Phone Number

Miami, Florida, United States

Incoming full-time 2025 on systematic options

Los Angeles, California, United States

NLP research @ DILL lab under Prof. Swabha Swayamdipta

New York City Metropolitan Area

- QR intern on high-frequency futures desk- Created general-purpose data sampler with hot-swappable filters and statistics to support microstructure and strategy research- Researched exchange channel latency variations across different order matching engines to significantly inform channel arbitration and interactable risk exposures- Conducted high-frequency strategy research (alpha sniffing) to boost existing alpha, created profitable proof-of-concept model

Urbana-Champaign Area

Urbana-Champaign Area

- Leading project to create deep learning framework to mimic institutional analysis, similar to the equity research outlined in a prime brokerage deal- Examining the strategies investment banks use to define price targets through literature review and designing a model to imitate said process- Aiming to write a peer-reviewed paper on our findings by the end of the semester.- Interviewing / evaluating future Quant applicants

Urbana-Champaign Area

- Built pairs trading algorithm

Urbana-Champaign Area

- Worked under Prof. Dong Wang @ NCSA- Created a lightweight methodology for debiasing large foundation models, like ChatGPT and CLIP- Implemented our findings into three classical debiasing approaches: prefix tuning, adapter tuning, and LORA- Planning to publish paper as second author and present at NAACL or CSCW 2024.

Boston, Massachusetts, United States

- Overhauled and automated DIVE-LD data pipeline- Created Timescale AWS EC2 instance with Athena backup to store mission and vehicle data, totaling 8+ billion rows- Built custom parsing tool to automatically parse raw logs into queryable format (data was previously not queryable)- Redesigned vehicle offload & upload, enabling up to 50% increase in upload speeds- Integrated database with Grafana, implemented multiple live dashboards for common use cases (post-mission analysis, fleet summaries, etc.)

Urbana-Champaign Area

- Worked under Dr. Ian S. Brooks @ CHI- Worked with the Center for Disease Control (CDC) to detect vaccine and COVID-19 misinformation on TikTok- Created OCR computer vision model for recognizing text in TikTok videos for misinformation analysis.- Presented progress weekly at Center for Health Informatics meetings and worked with team to brainstorm solutions for curating and identifying misinformation (e.g., sentiment analysis, content filtering, etc.).

Urbana-Champaign Area

- Worked under Prof. Volodymyr Kindratenko @ NCSA- Built ODQA framework for CS and ECE course material questions, thus creating an AI teaching assistant (pre-ChatGPT)- Leveraged GPT-3 for embedding resources, like textbooks and lecture notes. Allows for multi-media search with properly-cited results (e.g., highlights in lecture transcripts)- System searches across 11 separate models in parallel for retrieval, generation, moderation, and result ranking- Published live access to the model on HuggingFace for students to use

- Built end-to-end deep learning model (multi-layered LSTM, Tensorflow) for sales forecasting- Created server-to-model data pipeline using SQL and Pandas to dynamically pull data live from the WordPress server into the model for preprocessing. Implemented Augmented Dickey-Fuller operation to test data for stationarity. Developed automatic preprocessing functions to format and scale data for the model, employing loss-reduction techniques such as lags- Designed a multi-layer LSTM model to accurately forecast sales across a variety of products, utilizing Dropout and Gaussian Noise layers to prevent overfitting and increase consistency. Validation loss reached a consistent low of .0042 during testing, and .0053 during evaluation. Model can accurately predict future trends (based on a single lag, using its prediction as the next t-value) for up to ~30 days. Deployed an AWS EC2 server for model development and connected it to a Jupyter Labinstance through ssh. Met weekly with a senior developer to discuss progress and set milestones- Experimented with various solutions/strategies (both TensorFlow and PyTorch) to ensure maximum accuracy across a wide set of products, while maintaining model explainability

Managed and led robotics team.- Taught high schoolers the fundamentals of coding and robotics- Built robot eligible for the 2022 First Robotics Competition- Attended and judged elementary school science fairs- gave feedback on students’ projects to developtheir critical thinking and inquisition skills- Helped organize effective fundraisers to pay for parts- Doubled average concurrent club attendance year-over-year (~5 to ~10)

- Added subscription service in HTML with MailChimp integration to track and record newsletter interest- Developed automated newsletter service: automatically track most popular articles, and consolidatethem into one weekly newsletter sent to subscribers through MailChimp (unfinished)- Collaborated with other front end developers to unify ideas into one synthesized product

Master Of Science - Ms, Computer Science

Bachelor'S Degree, Computer Science + Economics

Activities and Societies: James Scholar Honors Program, Honors Living-Learning Community, Dean’s ListGPA: 3.96 / 4.0

High School Diploma

Activities and Societies: - Captain of Tennis team - Captain of SLS Robotics Team - Asian Culture Club - Recycling Club - 140+.