Building AI systems, backends & automation.
I'm a software engineer combining AI agents, backend development, and systems programming. I build production-grade platforms, LLM pipelines, and automation workflows - from financial data marketplaces to voice AI and custom game engines.
QuantPlace.org
A B2B Data-as-a-Service marketplace for institutional-grade trading datasets - order book L2/L3, OHLCV, sentiment, and AI training data. Vendors upload datasets; quants and AI researchers discover, preview, and purchase them with escrow-protected payments.
- FastAPI + async SQLAlchemy backend with Celery workers for background jobs
- Next.js 15 + React 19 frontend with full auth flow and vendor dashboard
- Per-buyer watermarking and stolen-upload detection via SHA-256 fingerprints
- MCP server for AI agents to search and evaluate datasets in-IDE
- Plisio crypto payments with 3-day escrow and auto-release
- Validation pipeline with Benford's Law, OHLC checks, and timestamp analysis
- AWS Rekognition moderation bot, Reddit social volume index pipeline
- Built-in backtester, data quality checker, correlation explorer, dataset diff tool
Model Context Protocol server for QuantPlace. Lets AI agents (Claude, Cursor, Windsurf) search datasets, preview samples, check metadata and evaluate vendors directly inside the IDE - no browser needed.
Personal prompt engineering toolkit. A collection of reusable prompt templates, evaluation utilities and multi-step reasoning patterns built during AI model training work at Outlier.ai.
C++ integration of LLM-powered AI companions into Unreal Engine. Custom plugin bridging UE5 actor lifecycle with real-time language model responses for NPC dialogue.
Custom game engine built from scratch in C++ on top of OpenGL. Includes a scene graph, entity-component system, shader pipeline, and basic physics.
Sep 2025
- Refined multi-step reasoning prompts to improve model accuracy on complex logic tasks.
- Reviewed and scored AI outputs using detailed rubrics, marking edge cases and failure modes.
- Architected an automated LLM evaluation framework in LangChain to benchmark model performance across reasoning chains.
present
- Built a scalable DaaS backend to ingest and serve high-frequency financial datasets.
- Implemented MCP integration reducing AI researcher integration time by 90%.
- Designed validation pipeline, per-buyer watermarking, and stolen-upload detection.
- Full-stack: FastAPI backend, Next.js frontend, Celery workers, Cloudflare R2, Docker.
- Engineered low-latency Voice AI infrastructure using Retell AI with real-time CRM sync.
- Designed n8n automated workflows for RAG-powered customer support bots.
- Orchestrated multi-API integrations reducing operating costs by 50%+ vs. human-led ops.
- Built PayPal subscription and invoicing systems for multiple client services.
Dec 2024
- Worked on a seasonal battle pass system with monetization features.
- Implemented thematic server events to boost player engagement and retention.
- Maintained and improved the live codebase, resolving player-reported issues.
Open to interesting projects, collaborations, and contract work.