Michael Opsha
Michael Opsha

I build automation and agentic tools that take manual steps out of everyday workflows.

Computer science and applied math at Stony Brook University. Based in New York, NY.

Selected work

BetLab 2026

A sports-betting research system. It prices lines at retail sportsbooks against the sharp market's no-vig prices, logs every pick to an append-only ledger before the game, and grades each one by closing line value. A worker on Railway runs the schedule unattended: overnight grading, morning lines, and a pre-game pass every 15 minutes that pulls the news, reprices, publishes and sends alerts.

Backtests showed closing lines beat my own ratings models, so a model gets zero weight until a backtest earns it some. A Next.js web app and a SwiftUI iPhone app read from one Supabase database. Built over about 2,000 commits with Claude Code as the engineering partner; I set the direction and made the product decisions.

BetLab's Games screen: tonight's college basketball, NBA and NHL games with start times and counts of priced markets.
Tonight's slate
A game page for Bills at Vikings: the market spread and total, the model's rating, a quarterback change, and written notes on what changed.
Game notes
A player prop priced at 55% with a fair price of minus 122 and a 1.3% edge, above a four-leg parlay compared across FanDuel, DraftKings and Novig.
Prop and parlay pricing
  • Python, pandas, scikit-learn, Next.js, TypeScript, SwiftUI, Supabase
  • ~90k lines; 1,020 Python, 229 web and 201 iPhone tests
  • 47 Postgres migrations with row-level security
  • NFL, college football, NBA, college basketball, NHL, soccer and UFC
Repository
Morning Briefing System 2026

A daily automation that pulls from calendar, tasks, habits, and project notes and composes a single prioritized morning brief in Obsidian. Designed and shipped solo — OAuth, scheduling, state reconciliation, test suite — using Claude Code as a pair-programming tool. Bidirectional reconciliation reads the prior day's checked items back into source files; atomic writes keep it safe against cloud-sync conflicts.

  • TypeScript, Node.js, Claude API, Google Calendar API
  • ~130-test Vitest suite
  • TypeScript strict mode
Repository
Time Tracker 2024

A Monday.com plug-in adding automation to the Time Tracker column. Custom webhooks and API integrations; user API keys stored in MongoDB.

  • Node.js, TypeScript, MongoDB
Skedulr 2023

A native iOS app for tracking coursework and assignments across user-entered courses.

  • Swift, SwiftUI

Experience

Technology Intern, Product Engineering

Deloitte Technology – US June 2026 – August 2026
  • Built a proof-of-concept agentic workflow in n8n with OpenAI models for a Government & Public Services team, querying USASpending.gov federal award data through natural-language prompts to replace manual contract lookup.
  • Project-managed a 5-intern team in the firm's Innovation Challenge; proposed retrieval-based conversation context management — retrieving relevant prior turns rather than resending full history — and modeled projected token-cost savings across model tiers, presenting to firm management.

Technology Intern

Deloitte Technology – US June 2025 – August 2025
  • Refactored a Python Gen-AI chatbot for the PxE Solutions team, cutting a ~2,000-line flow-control file to ~500 by delegating conversation control to OpenAI's LLM — reducing rigidity, bugs, and maintenance cost.
  • Rebuilt rejected Stage 1 deliverables into a working solution, adopted as the framework for Stage 2 development.
  • Built a Python scheduler plug-in with the NY CXO Support team for workflow management across 20+ technicians.

IT Support Staff

Community Resources Staten Island, Inc. April 2023 – January 2025
  • Provided technical support for 350+ staff across 19 locations; configured hardware/software and resolved maintenance tickets.
  • Deployed enterprise solutions including Cisco Meraki wireless access points and PaperCut print management.

Skills

Languages
Python, Java, JavaScript/TypeScript, Node.js, Swift, C, HTML/CSS
Tools
MongoDB, Docker, n8n, pandas, Git, Claude Code, Xcode
Spoken
English (native), Russian (fluent)

Education

B.S. Computer Science and Applied Math & Statistics

Stony Brook University Expected May 2027
  • GPA 3.50/4.00. Concentration in AI and Data Science.

Involvement

President, Stony Brook Men's Soccer Club

May 2025 – Present