CV
Education
- B.A. Computer Science, UC Berkeley — Expected 2027
Work & Research Experience
- Undergraduate Research Assistant, Sky Lab — UC Berkeley
- July 2025 – Present.
- Conducting research on multi-agent and AI systems, focusing on large language model function calling ability.
- Added local inference support for open-source models like Qwen3 and Kimi-K2-Instruct for the Gorilla LLM benchmarking framework.
- Designed multimodal evaluation tasks and reasoning-trace datasets to stress-test model performance.
- Exploring memory management for large ML systems, specifically policies to optimize KV-cache hit rates.
- Course Staff, CS161 – Computer Security — UC Berkeley EECS
- Summer 2025 – Present
- Lead weekly discussions and office hours for 400+ students, covering memory safety, cryptographic protocols, and secure system design.
- Debug student code, review security implementations, and grade assignments.
- Software Engineer, Admittere – AI-Powered College Application Assistant — Berkeley, CA
- June 2025 – Present
- Launched an AI assistant guiding students through Common App essays via a retrieval-augmented generation (RAG) pipeline.
- Designed a multi-tenant Supabase backend with pgvector similarity search and serverless functions for sub-100ms retrieval.
- Built a React + TypeScript front-end with OAuth authentication and Stripe billing.
- Software Engineer, realstudio.ai — Berkeley, CA
- Jan 2024 – Aug 2024
- Engineered a full-stack AI image generation platform with adjustable resolution and prompt control, scaling to 1,000+ concurrent users.
- Integrated LemonSqueezy for subscription billing and automated plan management.
- Fine-tuned Stable Diffusion v2 with DPO for realism and implemented Docker-based CI/CD pipelines.
Projects
- In-Simulated Self-Driving Car Model — Keras, CV (Mar–May 2025)
- Built a CNN to predict steering angles from camera images; applied normalization, cropping, and dropout, reducing MSE by 20%.
- Multi-Camera People Counting & Occupancy Analytics — OpenCV, YOLO, Flask (Mar–May 2025)
- Designed a real-time system with YOLO + SORT to count people via ID tracking and line-crossing; added ROI cropping, perspective correction, and smoothing to avoid double counts.
- Secure File Sharing System — Go, Cryptography (Feb–Apr 2025)
- Implemented hybrid encryption with MACs and digital signatures, ensuring authenticated access and resilience against database compromise.
- RISC-V CPU Simulator — Logisim (Aug–Dec 2024)
- Designed a pipelined simulator with hazard detection and forwarding for registers, memory, and control flow.
Technical Skills
- Programming Languages: Python, C, C++, Go, Java, TypeScript, SQL
- Machine Learning & AI: PyTorch, TensorFlow, Keras, Scikit-learn, CUDA programming, RAG pipelines, multimodal dataset design, Stable Diffusion fine-tuning
- Systems & Tools: Docker, Supabase (pgvector, RLS), PostgreSQL, Flask, NumPy, Pandas, React, Tailwind, CI/CD pipelines
Teaching
