Frankline Oyolo, Misango
I am a Quantitative developer with 3+ years of entrepreneurial and contractual experience building high-performance trading systems and financial models. My mission is to democratize access to low-latency markets free to developing world traders through optimized engineering practices.

Lumibot Algorithms
Live SPY Algorithmic Backed bots for Day Trading Stocks by utilizing ALPACA / IBrokers as the REST Agents.
Algorithmic Trading and HFT Research
Quantitative research and Engineering on various promising strategies, existing quant problems and opportunities using Mathematics and Statistics with end-to-end Data Science workflows.
Trading Terminal
A Quantitative Pre-Trading Research tool for stocks, Crypto, Govt Bonds and ETFs hosted on AWS EC2. Migration to JavaScript ongoing.
HFT FPGA Accelerator
FPGA, CuDNN/CUDA & C++ Hybrid infrastructure for ultra low-latency algorithmic trading. Implements market data parsing, order book management, and strategy logic in Verilog/VHDL. Designed for Xilinx boards.
Lync
A Segmented, thin and very light accelerator core implemented in SystemVerilog, featuring ITCH market data parsing, order book reconstruction, and sub-microsecond risk-check logic.
Fluxion
A GPU-accelerated toolkit for HFT analytics. CUDA/cuBLAS/cuDNN workflows applications to financial workloads that demand massive parallel math, high throughput, and near real-time modeling.
Embeddify
A simple app to help find the most relevant job faster using NLP, Typescript, Python, Docker and PostgreSQL.
x86 Router Exploits
Reverse engineering exploit script for mikrotik routers; blueprinting for other routers. Security research on x86 router vulnerabilities.
Data Structures & Algorithms
Scholarly and Self-tutelage on Intermediate and Advanced Data Structures and Algorithms in C, Java and Python.
Congressional Signals: A Systematic Strategy for Event-Driven Market Positioning
A research note exploring how congressional activity can be transformed into a systematic signal for market positioning and event-driven trading decisions.
Crypto Macro-Fundamental Strategy: A Research Note on Macro and Fundamental Signal Integration
A research note on combining macro and fundamental inputs into a systematic framework for cryptocurrency decision-making and positioning.
- • How to Hack CCTV Cameras in a Secured Network by Jamming WPA2/3 Exchange packets (Sep 2024)
- • How to Crack WPA2 WIFI Router Password (Feb 2023)
- • How to Remotely Exploit a Windows 10 PC's Webcam using Metasploit (Feb 2023)
- • Information Gathering Series; Part B: Basic Enumeration → Whois (Apr 2023)
- • Information Gathering Series; Part A: Recon-ng (Apr 2023)
- • How to Kick People Out of a WI-FI Network (Mar 2023)
- • Website Hacking Series; Part E: Cross Site Scripting(XSS) (Mar 2023)
- • Website Hacking Series; Part D: Cross-Site Request Forgery(CSRF) (Mar 2023)
- • Website Hacking Series; Part C: Manual & Automated SQL Injection (Mar 2023)
- • PicoCTF challenges: Easiest way to tackle Transformation: 104 (Feb 2023)
- • PicoCTF challenges: Easiest way to tackle Information: 168 (Feb 2023)
- • PicoCTF challenges: Easiest way to tackle NetCat: 168 (Feb 2023)
- • Website Hacking Series; Part B: Malicious Code injection & File Upload (Feb 2023)
- • Website Hacking Series; Part A: Bruteforcing using Burpsuite (Feb 2023)
- Dec 2025: Granted The All Builders Welcome Grant by Amazon to attend the AWS Re:Invent conference in Las Vegas
- Apr 2025: Granted The All Builders Welcome Grant by Amazon to attend the AWS Reinforce conference in Philadelphia
- Mar 2025: Granted the PyconUS Travel Grant by the Python Software Foundation to attend the PyCON in Pennsylvania
- Jan 2025: Awarded the Summer @ EPFL Grant of CHF 5000 of a 1.6% competitive rate globally to complete the SURF
- Dec 2024: Granted The All Builders Welcome Grant by Amazon to attend the AWS Re:Invent conference in Las Vegas
- Apr 2023: Awarded $5000 for winning the AI Tech4good Hackathon organized by Accenture
- Sep 2022: Awarded $10,000 as African Impact Grant Award for winning the Healthcare theme by Mastercard Foundation
- Aug 2022: Granted $150,000 for 4 years, termed Future Leaders Scholarship by HKU, to study Engineering
- May 2022: Granted CHF 4000 by Glencore to attend the Swiss-African Hydrogen case challenge in Switzerland
- • Citi — Markets Quantitative Analysis (MQA) Job Simulation — Forage (May 2026)
- • Citi — Markets Sales & Trading Job Simulation — Forage (Mar 2026)
- • AWS Certified Machine Learning – Specialty — Amazon Web Services (Jun 2025)
- • AWS Solutions Architect Professional — Amazon Web Services
- • Financial Modeling & Valuation Analyst (FMVA)® — Corporate Finance Institute® (CFI) (Dec 2022)
- • Android Developer Associate — Google (Mar 2023)
- • Student Member — Hong Kong Computer Society (Oct 2025)
- • Learn Ethical Hacking From A-Z: Beginner To Expert Course — Udemy (Feb 2022)
- • The Complete Android Ethical Hacking Practical Course C|AEHP — Udemy (Oct 2022)
- • Scientific Computing with Python — freeCodeCamp (Jan 2023)
- • Data Analysis with Python — freeCodeCamp (Mar 2023)
- • Machine Learning with Python — freeCodeCamp (Jan 2023)
- • Back End Development and APIs — freeCodeCamp (Mar 2023)
- • Front End Development Libraries — freeCodeCamp (Feb 2023)
- • Responsive Web Design — freeCodeCamp (Oct 2022)
- • Analysis of Algorithms (Mathematical) - Stanford University
- • Analysis of Algorithms (Practical) - Princeton University
- • Algorithms, Part I - Princeton University
- • Algorithms, Part II - Princeton University
- • Financial Markets - Yale University (Completed)
- • Analysis of Algorithms - Princeton University
- • Mathematics for Machine Learning - Imperial College London
- • Machine Learning Operations - Duke University
- • Machine Learning - University of Colorado Boulder
- • Portfolio Construction and Analysis with Python - EDHEC Business School
- • Internet of Things and Embedded Systems - University of California, Irvine
- • Advanced Data Structures, RSA and Quantum Algorithms - University of Colorado Boulder
- • FAKE: Fake Money, Fake Teachers, Fake Assets by Robert T. Kiyosaki (2019)
- • Freakonomics: A Rogue Economist Explores the Hidden Side of Everything by Steven D. Levitt and Stephen J. Dubner (2005)
- • Why the Rich Are Getting Richer by Robert T. Kiyosaki (2017)
- • Chip War: The Fight for the World's Most Critical Technology by Chris Miller (2022)
- • The Intelligent Investor: The Definitive Book on Value Investing by Benjamin Graham (1949)
- • Everything I Know About Love by Dolly Alderton (2018)
- • Think and Grow Rich by Napoleon Hill (1937)
- • Security Analysis by Benjamin Graham and David Dodd (1934)
- • Death: An Inside Story by Sadhguru (2020)
- • Illuminating Silence: The Practice of Chinese Zen by Master Sheng-yen and Dr. John H. Crook (2002)
- • Market Wizards: Interviews with Top Traders by Jack D. Schwager (1989)
- • The Selfish Gene by Richard Dawkins (1976)
- • 1776 by David McCullough (2005)
- • The Real Book of Real Estate: Real Experts. Real Stories. Real Life. by Robert T. Kiyosaki (2009)
- • Principles for Dealing with the Changing World Order: Why Nations Succeed and Fail by Ray Dalio (2021)
- • The Federalist Papers by Alexander Hamilton, James Madison, and John Jay (1787-1788)
- Engineered lightweight hardware accelerator cores in SystemVerilog featuring ultra-fast ITCH market data parsing, real-time order book reconstruction, and sub-microsecond risk-check logic.
- Developed bare-metal, deterministic operating systems optimized for high-frequency trading (HFT) written in C/ASM, utilizing custom thread scheduling to achieve nanosecond-level execution latency.
- Designed customized digital logic for high-frequency trading FPGA boards using Verilog and VHDL to process high-throughput live market feeds.
- Programmed LOB analytics engines in C++ to ingest and analyze order book imperfections, structural imbalances, and momentary pricing mishaps via low-latency broker websockets.
- Formulated multi-language trading frameworks leveraging Rust for production execution speed alongside Python for rapid alpha signal generation and automated circuit breaker deployment.
- Built a retrieval-augmented generation (RAG) system utilizing Semantic Kernel architecture in C# to automate quantitative research workflows by processing unstructured alternative data and market trends.
- Designed interactive global maritime data pipelines and alternative macroeconomic ingestion engines to map global trade flows and isolate predictive macro factors.
- Constructed a unified historical data-fetching pipeline and high-performance simulation engine in C# via Lean CLI with dynamic slippage and risk modeling.
- Fundraised proprietary trading capital alongside structured Limited Partnerships, managing small client advisory relationships while sustaining high annualized returns.
- Scaled a cross-functional team of quantitative researchers and engineers by strategically sourcing top STEM friends from elite global institutions.
- Ideated systematic multi-asset strategies across equities, FX, and crypto, strictly maintaining maximum drawdown under 8% via dynamic position sizing.
- Spearheaded a proprietary cross-asset correlation trading framework to capture structural relationship deviations, generating approximately 30% in annual alpha with a 1.5 Sharpe ratio.
- Enhanced traditional carry strategies using GARCH volatility forecasting and forward rate bias corrections to achieve atleast 20% annual returns.
- Directed the ingestion of alternative economic indicators, satellite data, and social sentiment into core investment models, boosting strategy performance.
- Authored QR research papers validating a statistical edge, predictability, and full reproducibility of the firm's potential proprietary strategies.
- Primary rotational class (Equities) • Multi-Asset Group (MAG) Structuring: Developing structured product solutions, cross-asset derivatives pricing, and quantitative payoff models for institutional and retail clients.
- Derivatives Trading: Analyzing market liquidity, managing risk parameters, and monitoring order execution workflows across equity derivatives and flow products.
- Architected RAG pipeline using GPT-4 Turbo (128k context) with Azure Cognitive Search vector indexing, achieving 93% query accuracy on BASF's TPM KB (58k+ entries) through hybrid semantic/keyword search optimization
- Developed multi-agent system with AutoGen Framework v1.2, implementing custom tooling for SAP TPM integration that reduced shipment cost analysis runtime from 14min to 38sec (22x improvement)
- Containerized Python analytics tools using Docker/Kubernetes on Azure Kubernetes Service (AKS), achieving 99.95% uptime through pod auto-scaling (HPAv2) and distributed Redis caching
- Optimized PySpark ETL jobs on Databricks (Delta Lake 3.1), processing 18TB weekly Inventory and Order data with 92% parallelization efficiency using optimized shuffle partitions (200+ core cluster)
- Implemented a genetic algorithm for route optimization using DEAP framework, reducing transportation costs by 31.4% on BASF's EU logistics network (4,200+ nodes)
- Engineered hybrid contextual generation pipeline using LangChain + Azure OpenAI GPT-4 Turbo, achieving 92.4% description accuracy across 312k Purview assets (38% error reduction vs manual)
- Implemented dynamic few-shot learning with Azure Cognitive Search index (1.2M reference docs) to cut manual metadata remediation by 650h/month
- Developed automated GPT-Lint test suite with PyTest + Azure ML, increasing compliance with 23 Manulife global standards (GDPR/CCPA) by 41% YoY. Reduced hallucination rates from 12.7% → 3.8% through: Bi-weekly prompt toxicity scans (OWASP LLM Top 10 aligned) and Output validation against SWIFT financial taxonomy (87.3% precision/recall)
- Architected 48 test scenarios for production GAI features using Azure: Stress-tested RAG pipelines at 12k QPS (SLA: 99.95% uptime) and Identified 17 critical path bugs pre-launch via synthetic data fuzzing
- Built Looker-Python dashboard (Plotly/Dash) analyzing 12K+ user feedback events → 75%↑ task completion rate via A/B-tested UI changes and NLTK sentiment tagging (89.2% accuracy)
- Automated GCP ETL with Prefect (20h→2h/week reporting)
- Scraped 8K+/day HK job ads (Scrapy/Selenium + 2CAPTCHA) → 16 sector-specific DataFrames
- Trained LightGBM model (AUC=0.91) for AI job matching using spaCy/BERT embeddings (93% user-interest recall)
- Prepared and facilitated APAC College presentations to an audience of 300 students
- Graded and prepared feedback on 50+ essay prompts from scholars and including practice resources to correct specific weaknesses
- Introduced & Trained 10 scholars to the entrance exams: SATs, DTE and directed them on extra resources where they can practice further
- Maintained records and progress reports for 10 scholars and flagged issues with the Study Group Leader for each respective country
- Assisted in the Selection of 300 scholars from Rwanda, DRC and Uganda to join the College Counselling 2023/24 class
- Selected for the prestigious 2025 Summer@EPFL Programme, with an acceptance rate of just 1.3% globally.
- Co-authored a peer-reviewed study proposing a Diffusion Convolutional Recurrent Neural Network (DCRNN) for high-resolution spatio-temporal forecasting of bike-sharing OD flows across Switzerland.
- Optimized distributed training workflows on the EPFL high-performance computing (HPC) cluster, utilizing multi-GPU parallelization and efficient data caching to maximize throughput.
- Designed and implemented a multi-scale feature integration framework, extracting OpenStreetMap (OSM) and population features at 500m, 1000m, and 1500m radii to encode hierarchical urban context for over 1,000 stations.
- Formulated the prediction task as a sequence-to-sequence learning problem, leveraging DCGRU cells to model spatial diffusion and temporal dependencies in mobility networks.
- Constructed a comprehensive dataset comprising 10M+ trip records, multi-scale OSM features, and population grids, enabling robust model training and evaluation.
- Demonstrated that the proposed DCRNN architecture achieves a 27% reduction in RMSE compared to ST-GCN and a 13.4% performance gain attributable to OSM feature integration, as shown by ablation studies.
- Developed attention-based interpretability analyses, revealing that urban stations predominantly attend to local features (500m), while peripheral stations emphasize regional connectivity (1500m).
- Built SAM (Segment Anything Model) pipeline processing 400+ daily images via DeepLab3+ (PyTorch), segmenting 1K+ architectural blueprints & HK public housing units to identify 12 key aeration features correlating with 23% airborne disease reduction (R²=0.81).
- Automated 3 Roboflow COCO training pipelines with Apache Airflow/Kubernetes, orchestrating 150 weekly inference tasks (98.7% SLA) while cutting GPU idle time by 37% via spot instance scheduling.
- Led academic-industry team (5 PhDs + HKU Architecture Dept) converting 8.7M Vector3D labels to ML-compatible formats (Blender/Python ETL), achieving 99.2% CAD-Mesh alignment accuracy for disease simulation models.
- Implemented BERT-Large (340M parameters) with custom tokenization pipeline for semantic analysis of 30k+ student responses, extracting 128-dimensional embeddings with 94.2% topic classification accuracy
- Developed Meta BART-based bidirectional transformer sentiment analysis system with attention visualization, achieving F1 score of 0.89 on 5-point sentiment scale across 17 pedagogical categories
- Engineered distributed data lake architecture (S3-compatible object storage with 12TB capacity) processing 1000+ survey datasets with automated ETL pipelines, enabling sub-second query response for longitudinal analytics
