Software Development Engineer, Intern – Platform Efficiency - TikTok - San Jose, CA
(2025-09 - 2026-03)
- Developed scalable microservices in Golang for bulk CSV processing, optimizing latency and resource use by 30%.
- Built test automation frameworks using Python and Selenium, integrated into CI/CD pipelines with Jenkins.
- Built a RAG-based TTAM Copilot serving 500+ DAU with 10K+ documents, with MySQL schema design and queries optimized to 50ms.
- Implemented Redis caching with a 90+% hit rate, reducing database load by 40% and improving throughput.
- Owned a large-scale file processing pipeline using Redis BitMap, MinIO (2TB+ storage), Apache Tika, and Kafka for asynchronous processing.
- Built a retrieval engine with Elasticsearch, combining vector similarity and BM25 with MySQL permission filtering.
Software Development Engineer - Soul Wallet - San Francisco, CA
(2025-01 - 2025-08)
- Built a multi-tenant credit scoring platform processing $10M+ GMV across 50K+ users, with PostgreSQL schema and ETL pipeline design.
- Ingested 500K+ events/day using Kafka stream processing for real-time analytics and financial aggregation.
- Developed SQL-based feature engineering pipelines with 200+ features using window functions and CTEs for ML.
- Curated 10M+ training records across 15+ tables, building an MLP model achieving 90+% accuracy.
- Built an anti-fraud engine with 150+ SQL-based rules in PostgreSQL, processing transactions under 100ms.
- Refactored microservices using Kotlin and gRPC, with connection pooling reducing DB latency by 75%.
AI Engineer - Heurist AI - Seattle, WA
(2024-01 - 2024-08)
- Shipped an AI agent using LangChain with OpenAI GPT-4 function calling, deployed on Twitter, Telegram, and Coinbase, serving 2K+ users.
- Built an LLM routing layer selecting between GPT-4 and GPT-3.5-Turbo based on prompt complexity, reducing inference costs by 30%.
- Implemented Redis caching with an 85% hit rate, reducing API calls by 60%.
- Built evaluation pipelines with user feedback loops, with ETL jobs in Python, Redis, and PostgreSQL.
- Created SQL analytics tracking 20+ performance metrics, optimizing query latency from 250ms to 80ms.
- Hosted services on AWS Lambda with API Gateway, handling 10K+ requests/hour and cutting infra cost by 35%.
