David Kahdian
Computer Engineering & Applied Mathematics @ UCLA
Education
University of California, Los Angeles
Computer Engineering B.S., Applied Mathematics B.S. | Los Angeles, CA
- GPA: 3.982
- Honor Societies: Corporate Chair, Tau Beta Pi; Candidate, Upsilon Pi Epsilon and Eta Kappa Nu
- Relevant Coursework: Software Construction, ODE's, PDE's, Game Theory, Analysis, Probability Theory
Technical Skills
Development
C/C++, Python, JavaScript(JS), TypeScript(TS), Java, SQL, Bash
Tools & Platforms
Linux, VMware, VSCode, Jupyter, MATLAB, LaTeX, Markdown
Libraries & Frameworks
PyTorch, pandas, NumPy, Matplotlib, Selenium
Problem Solving
Solved July 2025 Jane Street puzzle; preparing for 2025 Putnam exam
Experience
AI Algorithms Researcher
UCLA StarAI Lab | Los Angeles, CA
- Analyzing 12+ AI compilation languages for succinctness, and query and transform tractability
- Developing graphical Knowledge Compilation Map visualizer website summarizing 20+ existing papers
- Extending research knowledge of probabilistic circuit optimization techniques with planned publications
Risk Analysis Research Intern
UCLA Garrick Institute for Risk Sciences
- Saved 100+ hours of recurring data entry tasks by automating conversion of 4 proprietary data types
- Validated risk data for Diablo Canyon Nuclear Power Plant (provides 23% of California's carbon-free energy)
- Improved statistical models in Fault Trees and Bayesian Networks, merging into production codebase via GitLab
- Collaborated with 4 interns using C++, Selenium, Django, React, and CI/CD pipelines to streamline development
Data Analyst
Harman International (JBL) & UCLA Epicenter Co-op
- Built predictive model for 2025 digital marketing allocation in a team of 8, informing regional budget strategy
- Optimized revenue to 11x marketing spend across 12 regions with psychographic segmentation
- Built Excel-based quantitative model predicting 11% revenue increase with Lagrange multiplier techniques
Projects
Machine Learning Portfolio Optimizer
Python
PyTorch
Pandas
Jupyter
- Developed a neural network powered portfolio optimizer using Modern Portfolio Theory and deep learning
- Mined 135 quarters of forward-looking macroeconomic indicators to train PyTorch neural network
- Achieved 80% R² accuracy with 33 quarters of out-of-sample test data to prevent look-ahead bias
- Optimized stock-bond ratio recommendation, published model to Jupyter Notebook
Heston Monte Carlo Simulator
Python
NumPy
JavaScript
C
WebAssembly
- Created interactive web-based option pricing tool using Monte Carlo simulation techniques
- Utilized Milstein discretization for 1000-step time series analysis, with comparison to Black-Scholes
- Compiled C script into WebAssembly to 10x speed for Stochastic Differential Equation (SDE) computation
- Accelerated simulations to 1000+ per second with only 100MB RAM and no parallelism
BruinPlan
Svelte
TypeScript
Puppeteer
Cytoscape.js
Python
- Built web+mobile application enabling UCLA quarterly planning & prerequisite tracking
- Developed Puppeteer script to collect 14,000+ courses and 130 majors of data
- Created Python + Gemini API pipeline to convert raw data into JSON
- Utilized Cytoscape.js and topological sort to create graph-based planning interface
FactCheckSpeech
JavaScript
Selenium
OpenAI API
- Collaborated with three other students to create speech transcribing application
- Produced Chrome extension capable of translating and fact-checking 100+ languages of speech
- Awarded 2nd place in Congressional App Challenge by Congressman Adam Schiff