Expertise

Areas where I can help

My strongest fit is software engineering, with practical overlap in data workflows, applied AI, and technical evaluation. I like work where code, data, and real user workflows meet.

01Software 02Data 03AI / ML 04Evaluation

Areas of

Expertise

Software Engineering

Build reliable tools around real workflows: APIs, auth, persistence, forms, editors, previews, exports, and deployment.

Where I have used it

PrelegalMillenia VenturesDocument Intelligence Lab

Tools

TypeScript

Java

React

Next.js

Node.js

FastAPI

REST APIs

Python

SQLAlchemy

JWT / HttpOnly auth

password hashing

relational schema design

PostgreSQL

SQLite

Docker

Git

Jest

Playwright

Authenticated document workflow

Built legal-document flows with saved drafts, live Markdown preview, JWT sessions, and PDF export.

Review-state data systems

Designed PostgreSQL-backed workflows for extracted fields, corrections, review status, and dossier outputs.

Full-stack testing

Used Jest and Playwright to check editor behavior, template flows, saved-document workflows, and generation paths.

Data & Automation

Turn messy documents, spreadsheets, and business inputs into structured records people can review and reuse.

Where I have used it

Millenia VenturesDocument Intelligence LabEEG Eye-State Classification

Tools

Python

SQL

pandas

Streamlit

PostgreSQL

Excel/Sheets

JSON

Markdown

openpyxl

Docling

OCR / layout parsing

Document review automation

Automated review of 1,000+ founder/company files and reduced manual review from 5 hours to 20 minutes.

Structured export workflows

Built outputs that move cleanly between JSON, Markdown, Excel, Streamlit review screens, and stakeholder reports.

Signal and error analysis

Used chronological splits, feature extraction, and error review to handle noisy EEG signal data honestly.

Applied AI / Machine Learning

Use AI where it supports a workflow, then validate the output instead of blindly trusting it.

Where I have used it

PrelegalDocument Intelligence LabFlappy Bird NeuroevolutionEEG Eye-State ClassificationMillenia matching workflow

Tools

OpenRouter

LiteLLM

LangChain

RAG

OpenAI / Anthropic APIs

Structured JSON

OCR / layout parsing

Embeddings

Semantic matching

scikit-learn

NumPy

SciPy

matplotlib

AI-assisted field collection

Built conversational intake that extracts structured fields, merges them into state, and renders legal documents.

Document intelligence workflows

Used LLM/VLM review, table correction, layout-aware parsing, and structured extraction for document-heavy tasks.

Modeling and simulation

Built a neuroevolution simulator and an EEG analysis workflow with honest metrics, feature review, and reproducible runs.

Testing & Technical Evaluation

Check whether software, data, and models behave correctly before relying on the result.

Where I have used it

PrelegalFlappy Bird NeuroevolutionEEG Eye-State ClassificationDocument Intelligence Lab

Tools

Jest

Playwright

Docker

GitHub

unit/integration/E2E testing

regression checks

Confusion matrices

ROC/PR curves

End-to-end workflow checks

Tested full user paths for editor behavior, templates, saved documents, and generation workflows.

Reproducible experiment runs

Used fixed seeds, saved logs, and policy snapshots to make neuroevolution results easier to inspect.

Leakage-aware evaluation

Used chronological EEG splits, ROC/PR curves, error review, and anomaly detection to avoid inflated conclusions.