CV & Resume

Incoming Post-Trade Software Engineer (Quant Developer)

New York, NY

Technical Co-Founder

San Francisco, CA

• Led a startup focused on simplifying gift card liquidation & trading, securing over $50K in initial funding, receiving YC W25 and Techstars NYC final round interviews (top 1% of 10,000+ start up applicants), and signing LOIs with 3+ CEOs (McDonald's, Dunkin, PDQ).
• Developed an end-to-end solution encompassing a trading algorithm, API-based financial SaaS for corporate integration, JWT and SSO security, a custom ML retail recommendation engine, B2B data analytics capabilities, and a Stripe payment system.
• Built a full-stack app (Dart/JS frontend, Python Django/Firebase/AWS backend) with Docker, Redis, ensuring secure, scalable transactions.

NLP Group Research Assistant

Stanford, CA

• Developed RAG and long-context retrieval models at SAIL's NLP group, collaborating with advisors Professor Christopher Ré, and Azalia Mirhoseini (Hazy & Scaling Lab) on projects featured in publications such as Building Long-Context Retrieval Models with LoCo and M2-BERT.
• Optimized NLP and ML pipelines by leveraging hyperparameter tuning, SGD, LoRA, and CoT techniques with Hugging Face and PyTorch to enhance model performance and to build benchmarks and evaluations of existing retrieval (RAG and long-context) models.
• Engineered scalable pipelines on H-100/A-100 GPUs, GCP, AWS, and Stanford clusters for database integration.

Technical Research Assistant

Stanford, CA

• Developed LLM models to classify Russian media propaganda using log probability analysis and transformer-based models.
• Collaborating with Dr. Sergey Sanovich on a research paper focused on AI-driven disinformation detection and policy implications.

Software Research Intern

Stanford, CA

• Conducted advanced statistical analyses using R, Stata, and Python on 5,000+ survey entries.
• Fine-tuned LLMs for Supreme Court decision prediction, processing over 1,500 files and classifying cases by specific justices.
• Presented academic research to 10+ professors, contributing to two publications with Dr. Malhotra.

Consultant

Stanford, CA

• Conducted in-depth market research and competitive analysis for Azure, interviewing over 30 individuals and gathering 50+ survey responses.
• Collaborated with 5 Microsoft product managers to develop and test cloud solutions, providing 14 actionable suggestions to the Azure product team.

Software Engineer Intern

Allen, TX

• Developed RESTful APIs using Kotlin, Spring Boot, and PostgreSQL to manage user options in an Agile environment.
• Collaborated on cloud integrations using GCP, GitLab, and Jenkins.

Research and Technical Lead

Stanford, CA

• Led a team of over 25 students to review and analyze economic research papers and publish opinion pieces.
• Developed and managed a platform for economic discourse, inspiring hundreds of readers through impactful content.
• Stanford Economic Review website
• Example article: "Can Benjamin Keep His Throne as the King of Currency?"

Executive Director of Expansion

USA

• Raised over $175,000 to support children in need of educational assistance and trauma healing.
• Organized tournaments and developed public speaking skills for local middle school students.
• Be The Light Youth Association website

Co-Founder and Instructor

USA

• Offered speech and debate coaching to underprivileged students, leading to state and national victories.
• Raised $15,000 for Asian American education awareness.
• New Age Learning website

Managing Director

USA

• Launched America's largest Extemporaneous Speaking platform with 200,000+ visitors.
• Provided free resources and lessons for students entering debate competitions.
• The Extemper's Bible website

LegalEval Large Language Model: Understanding Legal Texts

• Created a custom PyTorch data loader, improving LLaMA2 model accuracy by 15%.
• Implemented model parallelism with Hugging Face Accelerate and DeepSpeed ZeRO 3, reducing training time by 30% on A100 GPUs.
• Executed legal text classification tasks for SemEval 2023 LegalEval, achieving top 10% performance.

ACM Convolutional Neural Network Bird Classifier

• Developed a PyTorch-based bird classification model with 87% accuracy on 3,113 images.
• Reduced overfitting by 20% through data augmentation techniques and GAN image generation.