Louis Miguel Bernal

Freelance · Remote

• Built AWS Bedrock LLM services over the warehouse: document intelligence parsing loan files, appraisals and financials at 97% field accuracy, a confidence-scored entity-resolution matcher reconciling hundreds of properties against a 14K-row regulatory registry, and a RAG copilot on the Snowflake semantic layer, cutting manual entry 80%.
• Architected and own the Airflow DAG ingestion layer across 9 enterprise source systems, including Hypha, NetSuite and iLevel, merging into tiered Snowflake raw schemas with source and load-timestamp lineage columns and moving 2M+ records/day at 99.9% uptime behind keyset-paged loaders and loud-failure schema-drift guards.
• Engineered a multi-layer dbt transformation backbone on Snowflake (staging, intermediate, export and mart layers, 350+ models), standardizing 120+ tables behind reusable data-quality and performance-metric test macros and a severity-tagged validation catalogue, cutting model build and QC time 60% and warehouse compute spend 35%.
• Delivered the finance and accounting pipelines the business runs on, rebuilding a portfolio valuation model to a 99.7% identity match against the reporting tape, shipping general-ledger ingestion with automated reconciliation checks, and automating an underwriting-to-asset-system sync with read-compare-write logic that never overwrites a human entry.

Float Infinity · Sydney, AU

• Architected an automated Azure telemetry and reporting platform ingesting real-time signals from Azure Resources, Policy, Patching, and Backup through REST API pipelines orchestrated on Airflow DAGs, cutting incident detection time by 50%, slashing reporting overhead by 70-80%, and sustaining 99.9% availability.
• Engineered production ELT on Azure, pairing Python/SQLAlchemy ingestion with modular, version-controlled dbt models using incremental materialization and unique-key deduplication, orchestrated end-to-end by Airflow DAGs across 4+ enterprise platforms, cutting data redundancy by 65%, tripling query performance, and lifting cloud governance visibility by 80%+.
• Built an ASX options trading analytics platform end to end: engineered a provider-agnostic market data ingestion service (Python, FastAPI) pulling OHLCV price history into PostgreSQL via idempotent upserts and watchlist sync, exposed through an authenticated REST API, and built the Next.js/React frontend stock list/detail pages with candlestick charting and watchlist management using optimistic UI updates backed by an integration test suite against a live PostgreSQL database and CI.
• Developed an end-to-end NLP/ML quantitative trading algorithm across 15+ instruments, training sentiment and macro-regime models on real-time market sentiment, macroeconomic indicators, institutional positioning flows, and options data. Engineered 50+ features across technical momentum, sentiment, and macroeconomic layers for 72% directional accuracy and a backtested Sharpe Ratio of 1.8.

PASIA · Procurement and Supply Institute of Asia

• Automated large-scale data preprocessing pipelines using Python (pandas, scikit-learn) and machine learning-based imputation, processing over 1 million procurement and contract records, reducing manual effort by 85%.
• Optimized SQL ETL pipelines for automated ingestion and transformation of high-volume procurement data, ensuring 99% consistency across departments and daily data refreshes for real-time insights.
• Developed advanced Power BI and Excel visualizations of contractor procurement data, transforming 1M+ contract records into insights that surfaced high-value savings and efficiency gains.

CSIT Program Council, DLSU-D

Led the programming and creatives committee, directing technical initiatives, event development, and creative direction, alongside an executive role in finance and public relations.

• Director Programming & Creatives
• Exec Dir Finance & PR

Freelance · Remote

• Designed and shipped 40+ responsive websites and web applications for small-business and academic clients using JavaScript, React, and Next.js, owning each project from requirements gathering through production deployment.
• Delivered Power BI and Excel dashboards for business and academic clients, turning raw data into decision-ready insights and cutting manual reporting time by ~40%.
• Engineered repeatable data-cleaning and EDA workflows in Python (Pandas) and Excel, standardizing ingestion across client datasets and reducing manual prep time by ~40%.

Multi-venue derivatives terminal built with Float Infinity: Binance, OKX and MEXC ingestion, a Gemma 4 + FinBERT intelligence stack, VaR / Kelly risk controls and a real-time circuit breaker.

YOLOv11-powered mobile app for real-time detection of LTO road markings with auditory feedback for drivers and pedestrians.

• 0.87 mAP@0.5
• 120ms / frame on Snapdragon 7 Gen 1 Inference
• 12 markings Classes

Institutional-grade S&P 500 forecasting platform powered by a 3-layer LSTM trained on 90+ years of historical data, with Monte Carlo path simulation up to 2000 stochastic paths.

• 61.4% Direction Acc
• 1.42 Sharpe (paper)
• up to 2000 MC Paths

Enterprise-grade RAG-powered gift recommender with triple-validation, FAISS vector store, Groq Llama 3.3 backbone, and zero-hallucination guardrails.

• 0 / 200 sample queries Hallucinations
• 740 ms p50 latency
• 0.82 Relevance@5

End-to-end XGBoost pipeline predicting vehicle CO₂ emissions with feature engineering, gain-based explainability, and an interactive dashboard.

• 0.95 R²
• 9.0 g CO₂/km MAE
• 13.0 g CO₂/km RMSE

Streamlit trading dashboard with real-time stock data, RSI / Bollinger / MACD indicators, and dynamic-programming max-profit optimization across a 10-year window.

• 6 Indicators
• 10 years Window
• unlimited Tickers

De La Salle University Dasmariñas

• 3.83 GPA / 4.0
• ×4 Dean's Lister