University of Wisconsin-Madison · Madison, Wisconsin, USA
B.S. Computer Science and Data Science, Minors in Economics and Business. GPA: 3.6. May 2028
Sai Sujit Punukollu
Madison, Wisconsin 53715 · [email protected] · LinkedIn · GitHub
Professional summary
Computer Science and Data Science student at UW-Madison with hands-on research experience in LLMs, RAG architectures, and reinforcement learning. Built production-deployed ML systems and forecasting pipelines across AI research and industry internships. Seeking AI/ML opportunities where deep technical foundations and applied modeling experience intersect.
Education
Experience
Student Intern, Customer and Technology Department
June 2026 – August 2026
Alshaya International Group Dubai, UAE
- Built a time series forecasting pipeline using Google's TimesFM to project sales for Primark UAE store launch and achieving results within a 5% error rate. Performed Monte Carlo simulations to quantify uncertainty bands around projected sales and designed a seasonal adjustment for Ramadan-driven shifts in shopping behavior.
- Analyzed the relationship between customer lifetime value and regional geopolitical conflict, identifying correlations between unrest indicators and long-term spending behavior and synthesized findings into a stakeholder pitch deck, translating analysis into retail and expansion strategy recommendations.
- Assisted in creating a recommenders model to create brand and category based recommendations for a 1B+ customers based on customer specific purchase patterns and broader geographical based purchase patterns. Segregated customers to control and treatment group based on all given indicators (gender, region and classified personae) to avoid potential biases for A/B testing.
AI Research Intern
June 2025 – August 2025
Summer of AI Laboratory (SAIL), CDIS and OpenAI Sponsored Program Madison, WI
- Built an AI-powered IT support assistant using RAG pipelines with LangChain and HuggingFace LLMs, enabling natural language search across 1,000 ingested institutional documents.
- Developed a document ingestion pipeline that scraped and parsed university documentation using BeautifulSoup.
- Implemented vector embeddings and semantic retrieval to enable context-aware responses across institutional documentation.
DoIT Help Desk Agent
April 2025 – Present
Department of Information Technology, UW-Madison Madison, WI
- Provide technical support for on- and off-campus users via phone, email, and chat across learning and productivity platforms.
- Administer NetID and MFA (Duo), perform remote troubleshooting using Bomgar, and manage tickets in Cherwell with call handling via Cisco Webex and Finesse.
Projects
Kaufman Inspired Trading Engine
- Implements five market conditions from Kaufman's Trading Systems and Methods: a regime classifier routes each symbol to trend- following (linear-regression slope), adaptive-MA (KAMA + dual-MA confirmation), or mean-reversion (RSI/stochastic/Bollinger vote, SMA-200 filtered), with price-shock and macro-headwind overlays that gate entries and tighten stops.
- Runs unattended on a Raspberry Pi cluster with a FinBERT news pipeline feeding the macro overlay; each session writes an audi table journal comparing live results to pre-committed backtest expectations, with every rule cited to its source page.
- Provided an API that mirrors current trades and positions. Live results: 18 sessions since 2026-07-06, strategy -3.40% against S&P 500 -0.57% and Nasdaq 100 -4.82%.
Awards & honors
- MadData Hackathon 2025, Honorable Mention
- Badgers for Good Design-a-thon 2026, Crowd Favorite
Skills
- ML and AI
- PyTorch, TensorFlow, Keras, Scikit-learn, LangChain, Hugging Face, Stable Baselines3
- Languages
- Python, SQL, R, Java, JavaScript, Bash, MATLAB
- Data Analysis and Visualization
- Pandas, NumPy, Matplotlib, Seaborn, Plotly, Excel
- Other
- Git/GitHub, API Integration, Web Scraping (BeautifulSoup)