Actively Updated
Personal Website
Full Stack Web Development & Hugo Migration
  • Originally built with HTML and CSS/SCSS from scratch.
  • Currently migrating to Hugo with PaperMod theme for better content management.
  • Features pastel purple aesthetic, timeline views, and responsive design.
Actively Updated
Bias to Action
Open-Source ML & CS Learning Resource
  • Comprehensive educational system with 85+ markdown files spanning mathematics, algorithms, ML, and deep learning.
  • Structured 20-week curriculum progressing from foundations to transformers.
  • Starts from first principles with cross-referenced topics linking prerequisites to related concepts.
  • Full-text searchable documentation with MathJax rendering for LaTeX equations.
  • Built with MkDocs, Obsidian-compatible, and hosted on GitHub Pages.
2026
The Curator Agent
A Deterministic Cognitive Pipeline for AI Research Curation
  • Architected a modular Level 1 Agent with deterministic cognitive pipelines to autonomously track, score, and deduplicate daily AI research papers from Hugging Face, reducing manual research time from 2 hours to 3 minutes.
  • Implemented three high-cohesion modules: PerceptionTools (robust API ingestion), CognitiveBrain (Qwen-2.5-14B inference with dual-stage Scientist/Critic reasoning), and PaperArchive (vector-based semantic deduplication).
  • Optimized for resource efficiency on AWS Graviton ARM64 CPU clusters using bfloat16 precision.
  • Engineered circuit breaker logic and Pydantic guardrails to transform probabilistic LLM outputs into deterministic structured data.
2025
Expedia GenAI Hackathon - 2nd Place Winner
Expedia Astronaut Expedition 2025
  • Developed a traveler-facing GenAI product prototype enhancing the vacation rental experience.
  • Leveraged OpenAI models and prompt engineering to summarize host-traveler conversations.
  • Recognized among top 2 of 30+ teams; proposal influenced internal product innovation strategies.
2025
Traveler Persona & Avatar Personalization Engine
Expedia Brainwave AI Hackathon
  • Designed a system using LLMs (OpenAI) to summarize traveler behavior into personas for personalization.
  • Integrated semantic filtering and RAG techniques to retrieve context from conversational and booking data.
  • Enabled downstream badge recommendation and ranking refinement based on persona-matching.
2022 - Present
Pricing ML Systems at Expedia
Production ML for Revenue Optimization
  • Led deep learning model developments contributing to +$237.5M annualized GBV uplift.
  • Built hotel-level ML pipeline for marketing budget allocation (+2.3% annual profit uplift).
  • Developed optimization toolkit using mixed-integer programming for pricing decisions at scale.
2021
Doordash Estimated Delivery Time
Machine Learning Project
  • Case study and business understanding
  • Data preprocessing and analysis
  • ML predictions with linear, Lasso, random forest and XGBoost including hyperparameter tuning
2017 - 2022
Q-Chem
Quantum Chemistry Software Development
  • Fully in charge of the RAS2-SF method development in Q-Chem 5.
  • Implemented 3 new features: Charge Transfer Calculation, Relativistic Effect, and Solvent Effect Calculations.
  • Facilitated experimental designs of solar cell and semiconductor materials using mathematical predictions.
2020
Interface between Q-Chem and Psi4
Automation Tool for Quantum Chemistry
  • Automates the creation of input files for Psi4 program based on Q-Chem files.
  • Organizes output and analyzes matrices after Psi4 output is generated.