Portfolio header banner

AI systems / quant tooling / data products

Tony Koo builds decision systems.

 

Data scientist and quantitative systems builder focused on agentic analytics, privacy-safe workflows, and financial engineering tools that turn ambiguous data into reviewable decisions.

View Projects Download Résumé
6.09s to 0.86s

APAC correlation analytics optimization

~90%

Approximate field-level questionnaire extraction accuracy on a sanitized project evaluation sample

FAISS + RAG

Semantic caching for lower-cost LLM pipelines

Proof before polish

Systems I can explain, defend, and improve.

The portfolio now leads with the work that matters most: AI workflows with audit trails, quant tools with measurable speedups, and public-safe reports that show how the systems fit together.

What I build

AI, data, and quant decision systems

Targeting data scientist, AI engineer, and quant engineering roles where models need production judgment.

Evidence

Metrics, methods, and boundaries

Speedups, extraction accuracy, leakage-aware evaluation, and public-safe reports with limitations stated.

Agentic analytics

Healthcare chatbot architecture

Multi-agent workflow for natural-language SQL, statistical review, charting, and human checkpoints.

Read case study
0.86s

Quant analytics latency

Reduced APAC correlation analytics runtime from 6.09 seconds through workflow and computation optimization.

~90%

Questionnaire field extraction

Built a computer-vision workflow for medical PDF and image questionnaires with reviewable output boundaries.

Approximate field-level accuracy on a project-scoped sanitized evaluation sample; not a production benchmark.
RAG

Lower-cost LLM pipelines

Designed Transformer, LangChain, and FAISS-backed caching paths for structured extraction and retrieval.

Visual evidence

Project pages that show the work, not just describe it.

Major case studies now pair written method notes with owned diagrams, slide-derived artifacts, short muted demos, and licensed research references where reuse is clear.

Agentic healthcare analytics visual with intent router, SQL agent, stats agent, and clinician review Agent workflow

Healthcare analytics with review gates

Hero visual, architecture diagram, HITL flow, ReAct/Reflexion research cards, and demo clip.

 Legal NLP visual with judgments, feature extraction, Legal-BERT, and case metrics Evaluation design

Legal NLP without leakage theater

Extraction architecture, HAN diagram, model-results artifact, leakage walkthrough, and field note.

 Self-RL and RAG extraction pipeline visual with FAISS cache and JSON output Systems cost

FAISS cache decisions with fallback logic

Semantic-cache evidence, Reflexion/ReAct paper cards, and a short cache decision demo.

 Yield curve modeling visual with an upward-sloping term structure Quant tooling

Yield curves to risk-aware tooling

Curve artifact, IBKR architecture visual, assumptions-to-risk demo, and conservative deployment boundary.
tony-koo.json
{
"name": "Tony Koo Ye Long",
"education": "National University of Singapore",
"degree": "BSc. Business Analytics (Computing)",
"specialisation": "Quantitative Finance, Agentic Analytics & Data Systems",
"graduation": "Jun 2026",
"knownLanguages": ["Python", "SQL", "Java", "R", "VBA", "JavaScript"],
"focus": ["Quantitative Development", "Agentic AI", "Financial Analytics", "Privacy-Safe Data Systems"],
"resume": " Resume_TonyKooYeLong.pdf ",
"github": " https://github.com/LMDlifers ",
"email": " tonykooyelong@u.nus.edu "
}

Experience

Current AI/Data Role

SKEZI - Data Scientist Intern

French NLP, API extraction, PII-safe text processing, topic detection, and portfolio-safe RAG/CV evidence.

Quant Engineering

Tradition Energy - Quantitative Developer Intern

Yield curves, derivatives tooling, Bloomberg/LSEG integrations, and APAC analytics runtime improvement from 6.09s to 0.86s.

Foundation

NUS Business Analytics + Finance Diploma

Business analytics, programming, econometrics, financial risk, client operations, and medical operations experience.

Work Experience

Featured Experience

SKEZI logo

SKEZI

Data Scientist Intern

French NLP, API extraction, PII-safe processing, and topic detection

  • Performed API-based data extraction to streamline information processing and improve workflow efficiency.
  • Cleaned and lemmatized French-language text with SpaCy, removing PII for regulatory compliance.
  • Applied BERT, TF-IDF, LLM workflows, and HDBSCAN clustering for topic detection.
  • Prototyped RAG, FAISS caching, and computer-vision questionnaire extraction workflows; the project-scoped workflow reached approximately 90% field-level extraction accuracy on a sanitized evaluation sample.
Python SpaCy BERT RAG LangChain FAISS Transformers Computer Vision HDBSCAN LLMs GDPR

Jul 25 - Current

Paris, France

Tradition Energy logo

Tradition Energy

Quantitative Developer Intern

Financial derivatives analytics and quantitative tooling

  • Engineered automated yield curve construction workflows using interpolation techniques to support high-precision pricing for complex financial derivatives.
  • Synthesised academic methodologies into bespoke quantitative models, bridging theoretical research and production-ready analytical tools.
  • Developed high-performance correlation analytics for APAC markets using Bloomberg and LSEG API integrations, reducing calculation time from 6.09s to 0.86s.
  • Supported new financial-product work by combining market data analysis, modelling, and client-facing spreadsheet tooling.
Python VBA Quantitative Finance Financial Analytics

May 24 - Aug 24

May 25 - Jul 25

Singapore

Earlier Experience

Military Medical Institute

Medical Technician

Singapore Armed Forces medical operations and patient-record discipline

  • Provided immediate medical assistance during emergencies and supported patient stabilization in high-pressure situations.
  • Maintained patient records and medical logs to support communication across medical teams.

Jul 21 - Aug 23

Singapore

Tradition Energy logo

Tradition Energy Singapore

Middle Office Analyst Intern

Client onboarding, trade operations, and cross-office coordination

  • Supported onboarding for 20+ international clients by streamlining existing SOPs.
  • Coordinated with trading desks, operations teams, and London HQ to support accurate high-volume trade execution.

Sep 20 - May 21

Singapore

Education

National University of Singapore

School of Computing

BSc. in Business Analytics (Computing)

  • University Engineering Scholarship
  • Sustainability Exchange Champion
  • Tembusu College
  • Grade: Honours with Merit
  • Graduated: Jun 2026

Relevant Coursework

  • Advanced Analytics with Big Data Technologies
  • Machine Learning
  • Quantitative Finance
  • Big Data & Machine Learning
  • Econometric Modelling
  • Blockchain Technology
  • Data Management and Visualisation
  • Probability and Statistics
  • Calculus & Linear Algebra
  • Programming Methodology I & II
  • Data Structures and Algorithms
  • Application Systems Development
Python R MySQL Java HTML CSS JavaScript Tableau AWS Figma VBA Git



Aug 23 - Jun 26

Singapore

Korea University

International Winter Campus

Participated in Korea University's Winter Campus academic and cultural immersion programme. Gained a strong foundation in core accounting principles and applied them to real-world business scenarios, demonstrating the ability to classify transactions and determine appropriate accounting treatments.


Relevant Coursework

  • Principles of Financial Accounting

Dec 24 - Jan 25

South Korea

PSL University (Paris)

Venture Creation (Entrepreneurship & Business)

Participated in an intensive venture creation program focused on starting a business and entrepreneurship. Co-created an AI wardrobe styling app and collaborated with a diverse team to validate the business model.

  • Best Venture Team Award: Awarded for the most feasible business idea and strong execution strategy.

Jul 24 - Dec 24

Paris, France

Temasek Polytechnic

School of Business

Accountancy & Finance (Finance specialisation)

  • Accountancy & Finance Interest Group Vice-President
  • School of Business Ambassadors EXCO
  • Cumulative GPA: Diploma with Merit
  • Status: Graduated

Relevant Coursework

  • Security Analysis & Portfolio Management
  • Risk Management
  • International/Business Finance
  • Fundamentals of Taxation
  • Financial Technology
  • Information systems & Financial Analysis
  • Business Law
  • Business Statistics
  • Cost & Management Accounting I & II
Accounting Tax Financial analysis Portfolio Management FinTech

Apr 18 - May 21

Singapore

Technical Expertise

Quantitative Finance

  • Stochastic Calculus & Ito's Lemma
  • Options Pricing (Black-Scholes, Binomial Models)
  • Fixed Income Analytics (Yield Curves, Duration, Convexity)
  • Risk Management (VaR, CVaR, Greeks)
  • Algorithmic Trading & Market Microstructure
  • Time Series Analysis (ARIMA, GARCH)

Machine Learning & AI

  • Deep Reinforcement Learning (DQN, PPO, A3C)
  • Natural Language Processing (RAG, LLMs, Zero-shot)
  • Supervised Learning (Regression, Classification, Ensemble)
  • Unsupervised Learning (Clustering, HDBSCAN, UMAP)
  • Neural Networks (TensorFlow, PyTorch)
  • Model Optimization & Hyperparameter Tuning

Data Engineering

  • Big Data Processing (Apache Spark, PySpark)
  • ETL Pipelines & Data Warehousing
  • SQL Optimization & Database Design
  • API Development (REST, FastAPI)
  • Async Programming (asyncio, concurrent.futures)
  • Vector Databases (FAISS, Pinecone)

Software Engineering

  • Python (NumPy, pandas, scikit-learn, Streamlit)
  • R (ggplot2, dplyr, quantmod)
  • Java (OOP, Data Structures, Algorithms)
  • JavaScript (React, Node.js, DOM manipulation)
  • AWS, Figma, Microsoft Excel, Word, and PowerPoint
  • Version Control (Git, GitHub)
Languages: English native, Mandarin intermediate, French basic Competitions: Deloitte, WireCard FinTech, YouCard, Sustainability Exchange Interests: running, squash, basketball, cooking

Featured Projects

Public-safe summaries structured for fast review: problem, build, evidence boundary, stack, and deeper case-study notes.

Project Browser

My Projects

1 / 1
Agentic healthcare analytics architecture

Agentic Healthcare Analytics Chatbot

Problem

Clinical and research users need auditable healthcare data workflows.

Method

Multi-agent analytics workflow with routing, SQL, statistics, charts, and human review.

Evidence

Sanitized capstone-backed summary with private data excluded.

Evaluation

Review SQL trace quality, answer usefulness, and human approval boundaries before production.

Limits

Public version excludes private data and does not claim clinical deployment.

Open project detail

Evidence Index

My Projects

A compact view of domain, implementation signal, stack, and public evidence boundaries.

Agentic Healthcare Analytics Chatbot Healthcare AI

Multi-agent analytics workflow for natural-language SQL, schema grounding, statistical review, chart generation, and human-in-the-loop auditability.

LangGraph, SQL, DuckDB, OMOP CDM, ClinicalBERT, FAISS, Responsible AI.

Case study
Judicial Analytics: Legal NLP Legal AI

Leakage-aware legal NLP pipeline that converts lengthy judgments into structured case-level features and evaluation-ready records.

Legal-BERT, XGBoost, HAN, JSON QA, target-leakage audits, feature engineering.

Case study
Self-RL & RAG Extraction Pipelines LLM Systems

RAG extraction architecture with Transformers, LangChain, FAISS caching, OCR review, and iterative self-reflection loops for structured output quality.

Python, reinforcement learning, RAG, LLMs, FAISS, asyncio.

Technical note
IBKR Mean-Reversion Trading Prototype Quant Engineering

Team project. Tony's public commits cover the Ornstein–Uhlenbeck strategy, mean-reversion speed and signal calculations, equity adaptation, and project documentation.

Python, ib_insync, pandas, NumPy, SciPy.

Team repository
RentLock Escrow dApp Web3 Product

Smart-contract rental deposit workflow covering escrow state transitions, ERC-20 deposits, dispute escalation, verifier logic, and frontend-contract integration.

Smart contracts, ERC-20, MetaMask, IPFS concepts, mechanism design.

Case study
Smoodee Sustainable Food Tech Startup Operations

Food-tech venture work spanning sustainable supply-chain planning, business analytics, product positioning, and web presence for customer acquisition.

Entrepreneurship, supply chain, web development, analytics, sustainability.

Website
Loan Default Analytics Pipeline Credit Risk

End-to-end credit-risk workflow covering ETL, feature engineering, model comparison, calibration, explainability, and business-facing risk communication.

Python, PySpark, Apache Spark, Tableau, MySQL, ETL pipelines.

Case study
FairTracker Event Analytics Platform Geospatial App

Full-stack event discovery application with mapping, geolocation, event persistence, filtering, authentication, and crowd analytics boundaries.

JavaScript, Google Maps API, Bootstrap, MySQL, REST APIs, geospatial data.

Public summary only
WelfareHome Social Impact Platform Care Services

Responsive full-stack platform for appointment booking, resource management, role-based access control, and caregiver-family communication flows.

Full-stack development, MySQL, Bootstrap, REST APIs, UX design.

Public summary only
Fixed Income Yield Curve Modeling Financial Engineering

Quantitative toolkit for bond pricing, yield curve construction, duration, convexity, scenario analysis, and spreadsheet-assisted fixed-income workflows.

Python, NumPy, SciPy, VBA, fixed income, quantitative finance.

Quant note
Black-Scholes Options Analytics Dashboard Derivatives Analytics

Interactive options analytics dashboard covering Black-Scholes pricing, Greeks, volatility sensitivity, payoff visualization, and scenario testing.

Python, pandas, NumPy, Streamlit, Plotly, stochastic calculus, options theory.

Public summary only
Agentic healthcare analytics architecture

Agentic Healthcare Analytics Chatbot

Problem: Clinical and research users need auditable ways to ask natural-language questions over complex healthcare data without losing traceability. Built: Designed a multi-agent analytics workflow for intent routing, SQL generation, feature engineering, statistical testing, charting, and human review checkpoints before results are returned. Evidence: Sanitized from capstone documentation only. Public summary excludes private partner references, reviewer details, personal identifiers, and raw data examples.
LangGraph SQL Agentic AI Healthcare Analytics Auditability Human-in-the-Loop

Read sanitized report

Judicial analytics NLP pipeline

Judicial Analytics: Legal NLP

Problem: Litigation risk assessment requires structured features from lengthy judgments while avoiding hindsight bias and target leakage. Built: Created a legal text analytics pipeline with feature extraction, temporal isolation, leakage checks, Legal-BERT embeddings, and model benchmarking against traditional ML baselines. Evidence: Public version describes architecture and methods only. Raw judgments, private metadata, and evaluator details are not published.
NLP Legal-BERT XGBoost Feature Engineering Leakage Control Model Evaluation

Read sanitized report

Self-RL and RAG extraction pipeline visual

Self-RL & RAG Extraction Pipelines

Problem: Convert unstructured PDF and questionnaire content into reliable structured outputs while controlling LLM latency and cost. Built: Architected RAG pipelines with Transformers, LangChain, and FAISS-backed semantic caching. Added computer-vision extraction and a data-efficient self-RL loop for iterative output improvement. Evidence: Resume-backed internship project. Public version avoids proprietary datasets, raw documents, contact details, and internal implementation screenshots.
Python Reinforcement Learning NLP RAG LLMs FAISS asyncio

Read technical note

IBKR mean-reversion research prototype workflow

IBKR Mean-Reversion Trading Prototype

Problem: Explore SPY mean reversion using historical IBKR data and a transparent Ornstein–Uhlenbeck signal model. Tony's contribution: Implemented and refined the Ornstein–Uhlenbeck strategy, mean-reversion speed and signal calculations, equity adaptation, historical-data flow, and project documentation in a two-contributor repository. Evidence boundary: The public code demonstrates a research prototype with IBKR connectivity; it does not substantiate live profitability, production deployment, or walk-forward results.
Python ib_insync pandas NumPy SciPy

Links: Team repository

RentLock escrow dApp workflow

RentLock: Smart Contract Escrow dApp

Problem: Rental deposits create trust and custody problems when funds are held off-chain by one party without transparent settlement rules. Built: Designed a decentralized escrow workflow with lease lifecycle states, ERC-20-style test tokens, verifier escalation, frontend-contract integration, and evidence anchoring as a future extension. Evidence: Sanitized from project report. Public card excludes teammate references, academic identifiers, screenshots with private accounts, and unredacted report pages.
Smart Contracts ERC-20 Escrow Logic MetaMask Frontend Integration

Read sanitized report

Smoodee Sustainability Startup

Smoodee - Sustainable Food Tech Startup

Problem: Reduce food waste by transforming blemished produce into value-added products while optimizing supply chain logistics. Built: Co-founded sustainability startup focused on circular economy principles. Developed supply chain optimization model, business analytics dashboard, and partnered with local farms and food banks. Built full-stack web platform for customer orders and inventory management. Evidence: Included as entrepreneurship and product-building work. Public summary avoids customer data, partner details, and unverifiable operating metrics.
Entrepreneurship Supply Chain Web Development Business Analytics Sustainability

Links: Website

Loan default analytics dashboard visual

Loan Default Analytics Pipeline

Problem: Estimate credit default risk from structured borrower and loan features while keeping the modelling process interpretable for business users. Built: Designed a big-data analytics workflow covering data cleaning, feature engineering, model training, model comparison, and communication of risk drivers through business-facing analysis. Evidence: The public site keeps only a sanitized project summary and does not link the raw PDF or private submission material.
Python PySpark Apache Spark Tableau MySQL Big Data ETL Pipelines

Read sanitized report

FairTracker event platform map interface visual

FairTracker: Event Management & Analytics Platform

Problem: Develop a geospatial event tracking application to help users discover and manage local events with real-time location services and crowd analytics. Built: Built full-stack web application with interactive mapping interface using Google Maps API. Implemented real-time geolocation tracking, event clustering algorithms, user authentication system, and MySQL database for event persistence. Features distance-based filtering and category organization. Evidence: Included as an application engineering project. Public summary avoids user data, deployment claims, and analytics numbers that are not backed by public logs.
JavaScript Google Maps API Bootstrap MySQL RESTful APIs Geospatial Data
WelfareHome operations platform dashboard visual

WelfareHome: Social Impact Web Platform

Problem: Build a comprehensive web platform for elderly care services, streamlining appointment booking, resource management, and communication between caregivers and families. Built: Developed full-stack web application with responsive design using Bootstrap. Implemented backend with RESTful APIs, MySQL database for user and appointment management, and secure authentication with role-based access control for different user types. Evidence: Presented as a social-impact application build. Public version avoids beneficiary data and removes unsupported production/user metrics.
Full-stack Development MySQL Bootstrap RESTful APIs UX Design
Yield Curve Construction Models

Fixed Income Analytics: Yield Curve Modeling

Problem: Implement multiple yield curve construction models (Nelson-Siegel, Svensson, Monotone Convex) to enhance fixed income pricing accuracy for institutional clients in energy derivatives markets. Built: Developed Python-based yield curve fitting algorithms using NumPy and SciPy for numerical optimization. Integrated VBA with Excel for seamless trading system integration. Implemented Nelson-Siegel and Svensson parametric models plus monotone convex interpolation for smooth curve generation. Evidence: Resume-backed quantitative development work. The public card keeps proprietary pricing details out while retaining the reproducible technical framing and the documented 6.09s to 0.86s optimization result.
Python NumPy SciPy VBA Financial Mathematics Quantitative Finance Fixed Income

Read quant note

Options Pricing Calculator Dashboard

Black-Scholes Options Analytics Dashboard

Problem: Build an interactive options pricing calculator with real-time Greeks visualization, enabling traders to analyze option strategies and risk exposure across multiple scenarios. Built: Developed Python-based options analytics tool using Black-Scholes model implementation with NumPy. Created interactive web dashboard with Streamlit, featuring Plotly for 3D Greeks surface visualization. Added Monte Carlo simulation capabilities for exotic options pricing. Evidence: Research-style quantitative finance prototype covering comprehensive Greeks calculation (Delta, Gamma, Vega, Theta, Rho), volatility smile analysis, and strategy comparison tools (straddles, strangles, butterflies) with scenario heatmaps.
Python NumPy pandas Streamlit Plotly Stochastic Calculus Options Theory

Testimonials

"Tony demonstrated a deep understanding of quantitative finance topics including bonds, mean-variance analysis, forwards, futures, and options, and connected classroom theory with practical quantitative development."

Quantitative Finance Lecturer

Academic Testimonial

National University of Singapore

"Tony contributed meaningfully from requirements gathering and design through prototype development, bringing a thoughtful, structured approach to a complex real-world analytical problem."

Healthcare Analytics Supervisor

Capstone Testimonial

Clinical Research Collaboration

"Tony laid the groundwork for an AI chatbot project by building a foundational RAG pipeline and an automated data-collection workflow, delivering practical and reliable contributions to AI development."

AI Internship Supervisor

Recommendation Letter

Paris AI Startup

Reading & Research Diet

I read across markets, geopolitics, behavioral science, AI systems, and history to sharpen how I think about incentives, risk, institutional change, and decision-making under uncertainty.

Systems Thinking Macro Risk Behavioral Finance AI Research Institutional Change

Bookshelf

Selected Influences

Book cover: On China by Henry Kissinger
Geopolitics & Institutions

On China

Henry Kissinger

A long-view study of statecraft, strategic patience, negotiation posture, and how historical memory shapes institutional decision-making.

Technical Writing & Research

Public-safe technical notes, case studies, and research-style papers written for recruiters, senior data scientists, and AI engineers.

Explore the AI Engineering Research guides and curated reading →

Technical notes

Implementation lessons on retrieval, caching, latency, cost, and guardrails.

Project case studies

Sanitized reports that show problem framing, architecture, trade-offs, and evidence boundaries.

Research-style papers

Structured writeups with abstract, method, evaluation design, limitations, and next steps.

Legal NLP evaluation pipeline preview

Leakage-Aware Evaluation in Legal NLP

Research-Style Paper • 2026 • Evaluation Design

Temporal splits, label leakage risks, Legal-BERT baselines, structured extraction checks, case-level assumptions, and limits.

Legal NLP Evaluation Leakage Control
Read Paper

Certifications & Credentials

CFA Foundations

CFA Institute

Finance credential and training

Quantitative Methods Fixed Income Derivatives

University Engineering Scholarship

National University of Singapore

Aug 2023

Merit-Based Leadership Academic Excellence

Sustainability Exchange Champion

National University of Singapore

Recognized exchange and sustainability initiative

Sustainability Entrepreneurship Leadership

Get in touch

Open to AI engineering, data science, and quant tooling conversations where the work needs to be explainable, measurable, and responsibly shipped.

Email GitHub Resume

© 2026 Tony Koo. All rights reserved.

Loading Portfolio...

Preparing a few systems-thinking notes...