Mark Frost

Data and analytics platform for multifamily real estate. Backend on the Rails core (ETL Back-End, then Dashboards & Reports team); closed ~1,400 Linear issues across ETL, reporting and AI.

• Built the in-product MCP server that lets Claude and ChatGPT query customers' portfolio data: OAuth flow, ~10 read-only tools over dashboards, widgets, transactions and the variable/formula catalog, per-client connector guide, usage logging, PII gating and versioned AI user agreements.
• Designed the AI Assistant backend: chat API with SSE streaming, model registry across OpenAI and Anthropic, Markdown serializers that make financial widgets readable for LLMs, follow-up suggestions and context-window accounting. Migrated the stack from langchainrb to ruby_llm.
• Shipped the AI Smart Widget: LLM-written narrative summaries inside scheduled PDF reports, with prompt management, provider settings, preview and mock modes.
• Owned the backend of the Reporting Module (2024-2026): portfolio and asset-level PDF templates, Excel templates, report groups with recurring schedules, email and investor-portal distribution, cover pages, combined PDFs (HexaPDF) and bulk actions. Picked Gotenberg for rendering after an R&D spike.
• Built the Data Connections framework: inbound-email and SFTP connectors, manual uploads, parsing status and reparse flows, plus parsers and importers for Yardi, RealPage, Entrata, ResMan, AppFolio, MRI, QuickBooks, NetSuite, Sage, CYMA and RentEGI.
• Re-architected operational data around legal-entity management periods and LE-specific mappings (unit types, statuses, charge codes) across all ERPs, with SQL-view readers and zero-downtime widget migrations.
• Built the REconcile construction-finance module: change orders, invoices with approval thresholds and e-signatures, requisitions, draws, funding sources and payment emails. OpenAPI-first with Stoplight, RSWAG and Committee contract tests.
• Integrated a Python ML service for GL and charge-code mapping suggestions: Ruby workers and services on the Rails side, model retraining and dead-code cleanup on the Python side.
• Platform work: Rails 7.1 upgrade, per-customer Sidekiq queues, OpenAPI validation and linters in CI, Sentry-driven triage of production issues.

Resume builder with 20M+ users worldwide. Worked on core backend features and designed application architecture.

Led the backend team; maintained and developed new features for goodok.mts.ru, the MTS ringback-tone service.

Backend development on Ruby on Rails.

Backend for VR simulators with an IoT component: Ruby on Rails, RSpec, Docker, MQTT, WebSockets, MariaDB.

Developed Ruby on Rails applications.

MyIncome.ge

Income tracker for individual entrepreneurs in Georgia: currency conversion at NBG rates and 1% small business tax declaration prep.

Void VPN

Fast, private VPN delivered through a Telegram bot.