

Built an OOP-based configuration framework in Python using Pydantic to eliminate .env duplication across services and environments. The system centralizes configs into layered, reusable classes with environment-, service-, and instance-level overrides. Integrated with a CLI tool (ac) for seamless use in local dev and CI/CD, enabling type-safe, testable, and DRY configuration management.

Led the development of a robust CI/CD pipeline using GitHub Actions, enhancing deployment speed, security, and reliability. The pipeline automated secret scanning, vulnerability checks, testing, semantic versioning, and pre-release deployment, resulting in safer and faster delivery to production with traceable, reproducible builds.

Designed and implemented a resilient, queue-driven API gateway architecture that reduced downtime by 70% during peak traffic. The system used asynchronous processing, intelligent caching, and rate limiting to eliminate cascading failures and ensure stable performance under high load (40+ req/s).

Designed and built a scalable machine learning framework using monadic design principles to unify the research-to-production workflow. The system enabled composable, reproducible, and resilient pipelines, reducing development time by 50% and eliminating critical production errors across various ML tasks.

Implemented a Markov Chain-based risk labeling system to enhance credit risk modeling and portfolio analysis. The solution converts delayed binary labels into dynamic, continuous probability scores, improving model performance, enabling faster risk insights, and standardizing analysis across varying data histories.

Developed and benchmarked hybrid CNN-LSTM and LSTM-CNN models to forecast crude oil prices and generate trading signals. The AI-based strategies outperformed traditional methods like RSI and Buy & Hold, achieving up to 17.8% annual ROI and higher risk-adjusted returns (Sharpe ratio). The project highlights the power of deep learning in volatile, non-linear markets like oil trading.

Built a Python-based pipeline network modeling tool that automated data handling and scenario analysis for Chevron’s offshore operations. By representing pipeline networks as tree structures and integrating with simulation software, the tool saved $49.8M in capital costs and drastically reduced engineering workload.

Developed and deployed a linear programming-based optimization system at Chevron that increased oil production by 15.2%, generating $15.3M in annual revenue. The solution automated complex well operation decisions across 1,000 wells using real-time data, reducing human error and improving operational efficiency.

Built a modern, responsive portfolio website featuring dark/light theme switching, custom animations, and optimized performance. Implemented automated CI/CD deployment with testing coverage, demonstrating full-stack development capabilities from design to production.

ABACUS digital
: a fintech startup and subsidiary of SCB
Jul 2024 - Present
• Managed scalable data platforms, ETL pipelines, data warehousing, and microservices leveraging Python, Go, SQL, GCP services (e.g., BigQuery, Beam/Dataflow, Pub/Sub, Cloud SQL, Cloud Storage, Cloud Run, Firestore, Cloud Tasks), AWS, and Airflow.
• Developed private Python packages to enhance team productivity, including an OOP-based configs framework enabling seamless multi-environment releases and reusable CI/CD pipelines across projects; versioning automation; standardized logging; Dataflow abstractions; LangChain/LLM interfaces; and utilities for encryption, GCP services, Elasticsearch and more.
• Applied DevOps best practices with CI/CD and IaC tools (GitHub Actions, Terraform, Bash, Linux, Docker), reducing lead time, failure rate and time to recover while increasing deployment frequency.
• Engineered scalable and reliable real-time services, including API gateways built with FastAPI and asynchronous processing handling 40 req/s at peak to achieve ~70% downtime reduction and prevent cascading failures, and a real-time analytics service for approval model quota tracking, using GCP services, and Elasticsearch.
• Prototyped an LLM-powered data discovery tool using LangChain, embeddings, and vector search to enable natural language interaction with databases and NLP-based retrieval of relevant SQL from a centralized query bank.
• Provided leadership by mentoring junior engineers and interns, supporting team members, and fostering collaboration to enhance technical skills and drive project success.
• Drove software development excellence through rigorous code reviews, system design, architecture reviews, and test-driven development with automated unit, integration, and end-to-end tests, ensuring scalable, maintainable, and high-quality solutions for data engineering and AI/ML projects.

Honest Technology
: a fintech start-up with top-tier tech standards and DevOps practices.
Jul 2022 - Mar 2024
• Managed the production-grade ML development lifecycle (design, build, research, deploy, MLOps), contributed to hiring decisions, and fostered team growth.
• Spearheaded adoption of Markov chain for the credit risk model labeling, overcoming small and short-duration dataset limitations, improving AUC by 7%, and contributing to a $20 million funding round.
• Developed a scalable Python machine learning framework using monadic patterns, enabling reproducible research-to-production workflows, reducing development time by over 50%, and achieving zero critical incidents.
• Cultivated secure, scalable, maintainable, and reliable codebases by adhering to strict best practices across DevOps, IaC, CI/CD, testing, software development, coding standards, and security/vulnerability—using tools such as Terraform Cloud, GitHub Actions, SonarCloud, and Trivy.

(with Software Development experience)
Chevron
: a world-leading oil and gas company
Jul 2017 - Apr 2022
• Designed and developed software solutions using Python. One leveraged linear programming for production optimization, resulting in a 15.2% increase in oil production and $15.3 million/year revenue gain. Another enabled scalable pipeline network modeling, reducing investment costs by $49.8 million.
• Demonstrated exceptional project management skills in coordinating cross-functional teams for various operational projects driving $128.5 million more benefits and zero injury/incident from my projects.
(Merit scholarship from Chevron)
(1st Hons, Top 1 of the class)
React • Rust • Linux and Bash Scripting • GCP Cloud Engineer • Google Cloud Certified Data Engineer • Go • Scala • Deployment of Machine Learning Models • Langchain • Machine Learning Specialization by Stanford University • CS50x by Harvard University • MLOps Specialization by Deeplearning.ai • SQL & Databases • Fundamentals of Database Engineering • Google Cloud Digital Leader Training • Clean Code • PySpark • DevOps: Infrastructure as Code With Terraform • Docker & Kubernetes • Feature Engineering • Feature Selection • Hyperparameter Optimization • Machine Learning & Data Science Bootcamp • System Design • Design Patterns • Functional Programming • Python & Golang • Python OOP • Git & Github • Machine Learning Expert • The Complete Web Developer • Frontend Expert • Algorithm & Data Structure • Data Scientist with Python Track by Data Camp
Chartered Financial Analyst (CFA) I • TOEFL ITP (560/677) • TOEIC (925/990)