PROJECTS

Project Showcase

Selected engineering projects across strategy tooling, machine learning, accounting research, compiler design, and SEC data infrastructure.

PORTFOLIO / PROJECTS
Demo Product PreviewDORA
Verified Product Flow
Figma mobile preview
React dashboard routes
Strategy parser
Backtest display
Record
DoraHacks
Preview
Figma demo
Core
Parser
01 / BACKTEST PARSER2025HACK CANADA

Quantify

A Hack Canada project centered on parsing user-defined backtesting strategies and displaying the resulting strategy/backtest workflow.

Problem

Backtesting logic is hard to explain clearly when strategy conditions, parsed rules, and result displays live in separate pieces.

Impact

The project turns strategy inputs into a clearer workflow: build the rule expression, submit it to the parser, and review the backtest display.

Stack
ReactFlaskMySQLPythonChart.jsStrategy Parser

Visual is pulled from the DoraHacks-linked teammate Figma/demo preview; copy and stack were checked against the public Quantify GitHub repo.

Cyclica Prediction Model ROC curve and model diagnostics
Model Evidence
AlphaFold2 featuresVERIFIED
Binding classificationVERIFIED
ROC validationVERIFIED
Hackathon winnerVERIFIED
Signal
ROC AUC
Domain
Protein ML
Result
Winner
02 / AI / MACHINE LEARNING2023WINNER

Cyclica Prediction Model

Award-winning protein binding prediction project for Cyclica, built with feature engineering and machine learning.

Problem

The challenge was to predict drug-binding behavior from AlphaFold2 protein data with enough signal quality to be useful in a hackathon setting.

Impact

The model reached strong classification performance and won recognition in the Cyclica challenge.

Stack
PythonScikit-learnTensorFlowPandasBioPythonMatplotlib

Devpost page reviewed for the project title, challenge context, and winner status.

Private Source Project

This compiler project is intentionally not represented with source screenshots. The public page summarizes the implementation scope and keeps the source available upon request.

Pipeline
5 stages
Language
WLP4
Output
Assembly
Compilation Pipeline
Lexical analysis
Parsing
Semantic analysis
IR generation
Assembly output
03 / SYSTEMS / COMPILER2024PRIVATE

WLP4 Compiler

A C++ compiler with lexical analysis, parsing, semantic analysis, code generation, and assembly output.

Problem

The project turns a constrained language into lower-level code through a full compiler pipeline rather than a single parser demo.

Impact

It demonstrates systems fundamentals: formal grammars, type checking, symbol tables, and deterministic code generation.

Stack
C++MIPS AssemblyCompiler DesignParsingSemantic Analysis

No public screenshot or repository is shown for this private compiler project, so the visual stays at project-scope level.

Discuss Project
Reconciliation WorkbookANALYST REVIEW
Non-GAAP Reporting Analysis extracted reconciliation spreadsheet
Disclosure
Press release + filings
Adjustment
Mapped to account
Output
Research workbook
Extraction Flow
01
Filing
02
Parse
03
Normalize
04
Review
04 / ACCOUNTING RESEARCH2024RESEARCH

Non-GAAP Reporting Analysis

Research project collecting Non-GAAP metrics, reconciliation accounts, and management justifications from filings and releases.

Problem

Non-GAAP disclosures are buried in inconsistent tables and text, making scaled comparison hard for accounting researchers.

Impact

The demo shows how scraped filing evidence can be structured into analyst-ready reconciliation data for later review.

Stack
PythonNLPBeautifulSoupPandasExcelSEC Filings

GitHub repository kept as the main source link; page visual uses the existing research output screenshot.

Filing Index Schema
companyindexed
filingindexed
documentindexed
extraction_jobindexed
SELECT filings
WHERE form IN ('10-K', '8-K')
ORDER BY accepted_at DESC
SEC IDX EvidenceQUERYABLE
SEC Filings Database System SEC filing index spreadsheet
05 / DATA ENGINEERING2024RESEARCH

SEC Filings Database System

Customizable SEC filings database infrastructure for researchers who need fast access to filing metadata and source documents.

Problem

Researchers need WRDS-like access patterns, but with more control over filing types, source documents, and extraction needs.

Impact

The system turns raw SEC index data into a queryable research substrate for filing discovery and downstream extraction.

Stack
PythonPostgreSQLDjangoAWSDockerSEC EDGAR

GitHub repository kept as the main source link; page visual uses the existing SEC index screenshot.

Evidence-led visuals

Screenshots are framed as product evidence, not dumped raw.

Source-aware copy

Links now support the story each project is trying to tell.

Consistent layout

Every case study uses the same metadata, media, stack, and CTA system.