+12.3%
quinella betting ROI
Resume / ai-engineer
Henry Li
Senior AI Engineer
Staff AI Engineer and boutique software studio co-owner with 5+ years building production ML systems across consulting, logistics, and real-time platforms. Interviewed and iterated on nine model architectures — from LSTM-Transformer hybrids to Set Transformers to gradient-boosted ensembles — for structured prediction problems. Developed a quinella betting strategy achieving positive ROI through feature engineering, deep ensembles, and convex portfolio optimization. Owns the end-to-end ML lifecycle: problem framing, data pipelines, model development, evaluation, deployment, and monitoring.
9
model architectures explored
500/s
inference API throughput
<50ms
p99 prediction latency
capability map
Languages & Tools
PythonTypeScriptJavaSQLBashGit
ML & Deep Learning
PyTorchLightGBMXGBoostCatBoostscikit-learnNumPyPandasfeature engineeringhyperparameter tuningcross-validation
Model Architectures
LSTMTransformerSet Transformerdeep ensemblesattention mechanismsgradient-boosted treesranking models (λRank)
MLOps & Infrastructure
DuckDBFastAPIPostgreSQLDockerAWSLambdaSageMakerCI/CDGitHub Actions
Optimization & Evaluation
Kelly criterionconvex optimization (cvxpy)walk-forward validationportfolio optimizationrisk managementA/B testingstatistical testing
Data Engineering
ETL pipelinesweb scraping (Playwright)data modelingDuckDBPandas
delivery timeline
Staff AI Engineer
Horizon Technologies Advertisement
Jul 2025 - Present
- Co-own a boutique software studio and lead AI/ML-driven product delivery, building predictive models and intelligent automation for early-stage products and business teams across APAC.
- Own the full ML lifecycle — problem framing, data pipeline design, model development, evaluation, and deployment — translating ambiguous business goals into production AI systems.
- Designed and deployed a real-time prediction API for a Hong Kong-based fintech client, serving 500+ requests/second with sub-100ms p99 latency using an ensemble of gradient-boosted models on AWS Lambda and SageMaker.
- Established lean MLOps practices, reusable training pipelines, and reproducible evaluation frameworks that accelerate model iteration across client engagements.
Application Engineer
Arrow Electronics
Jun 2024 - Jun 2025
- Built and maintained data pipelines and ML-adjacent services for a warehouse management platform handling 1M+ daily transactions with 99.95% uptime requirements across global logistics hubs.
- Designed feature engineering workflows and data transformation layers that fed downstream predictive models for logistics demand forecasting and inventory optimization.
- Built CI/CD pipelines and Kubernetes deployment workflows that cut release lead time from 2 weeks to 2 days.
Senior Software Engineer
Amtran International
Oct 2022 - Apr 2024
- Led ML model development for a high-traffic platform serving 3,000+ concurrent users, iterating on model architectures from LSTM-based sequence models to gradient-boosted decision trees.
- Designed and validated feature engineering pipelines combining historical sequence data with static attributes, achieving 35% improvement in transaction-level efficiency.
- Established walk-forward validation, statistical testing, and ensemble evaluation practices that supported zero critical findings across two external audits.
System Analyst
Ace Technologies Limited
Apr 2022 - Sep 2022
- Designed data collection pipelines and built lightweight predictive features into a cross-platform app that maintained 60 FPS and supported $50k+/month in in-app revenue.
- Built server-side data aggregation and feature computation services feeding model-driven user engagement optimizations and personalization.
Analyst Programmer
Global Logistics System (HK) Limited
Nov 2021 - Apr 2022
- Built and maintained data processing pipelines and analytics workflows for a global logistics platform processing terabytes of freight data under 24/7 availability requirements.
- Delivered the Cathay Pacific Fly Greener carbon offset portal, supporting a 3x increase in corporate participation after launch.
Web Developer
MADTec Solutions Limited
Jan 2021 - Aug 2021
- Delivered secure, data-driven web applications for government and financial clients that passed penetration testing requirements.
- Supported a zero-downtime cloud migration that maintained 99.9% service levels and reduced hosting costs by 35%.
additional information
Languages
English (Fluent), Cantonese (Native), Mandarin (Proficient)
Education
BA Economics, The University of Manchester