BridgewaterHello there · AI automation · workflow systems
I build AI workflows that fix messy operations.
I’m Mike “Batman” Cohen, a business owner and self-taught automation builder who turns manual processes, broken data, and disconnected tools into systems people can actually use.
I ship systems, not demos.
My work is built around practical outcomes: saved review time, recovered data, cleaner handoffs, tighter feedback loops, and workflows that scale past one person holding everything together.
Business owner
Workflow builder
AI Operator
Startup advisor
Trusted environments
Notable clients and advisory companies
I’ve worked around high-expectation teams where workflows need to be clear, reliable, and useful to the people operating them.
Notable clients
Bridgewater
D. E. Shaw
PwC
Uber
Ross Stores
Whole Foods
GameStop
Waste Management
The New York Times
Lightmatter
SecurityScorecard
Redis Labs
Mattermost
phData
TeamSnap
Relativity SpaceAdvisory companies
SourceWhale
SeekOut
Clay
HoneIt
Pin
Sense
Loxo
Chatterworks
SmartRankSelected work
AI and automation systems I’ve built
Each project started with a messy workflow: manual review, broken data, repetitive setup, disconnected tools, or unreliable technical operations.
Build receipts
Code links included below for the featured builds
AI evaluation system
Candidate Scorer
Built a cloud-based AI application that evaluates profile data against role-specific criteria, calibrates from user feedback, and keeps each role self-contained while following the same repeatable workflow.
- Objective
- Automate the manual process of checking every profile returned in a search.
- Method
- CSV and Google Drive inputs, rubric-based scoring, calibration feedback, and role-specific artifacts.
- Result
- About 30 minutes of setup per role, automating roughly 80% of profile assessment and saving 3.5–6.5 hours per role.
- Tools
- Google Sheets, Clay, Google Apps Script, Perplexity Computer.
AI sourcing review app
Machine Learning Sourcing Automation App
Built a SQLite-backed candidate sourcing review surface for a Machine Learning Solutions Architect search, turning sourced profiles into a reviewer-friendly app with scoring logic, filters, calibration, and candidate detail review.
Outcome: recurring Brazil and US sourcing batches flow into a persistent review UI with ratings, feedback themes, review states, workflow settings, and reviewer notes.
Express · SQLite · vanilla JavaScript · Python pipeline · candidate scoring · sourcing review UI
Candidate matching engine
Candidate-to-department matching
Built a recruiter-operator tool that takes LinkedIn-style candidate CSVs, checks them against an embedded role library, and routes each profile to the best-fit department and role.
Outcome: template-ready CSV exports with confidence, rationale, department fit, and role fit appended to every candidate row.
TypeScript · React · Express · deterministic matcher · CSV export
Mobile-first PWA
Cute Translator App For Wife
Built a tiny, mobile-first Progressive Web App for everyday English ↔ Mexican Spanish text-message translation, with automatic language detection, one-tap copying, and iPhone Home Screen support.
Outcome: a personal-use translation tool that feels warm, fast, and useful without requiring authentication, a backend, a database, or a complex deployment pipeline.
HTML · CSS · JavaScript · PWA · Service Worker · MyMemory Translation API
Data workflows
Google Sheets automations
Created advanced Sheets, REGEX, Query, LLM, and Apps Script workflows to gather profile data, normalize ingestion across platforms, maintain hygiene, and automate ETL steps.
Outcome: one of the most detailed and efficient sourcing data workflows I’ve built.
Google Sheets · Llama LLM via Poe · Apps Script · Quickli
Data recovery agent
CRM data recovery automation
Built agents to recover roughly 6,000 candidate profiles, campaign history, metrics, and CRM mappings after a migration caused critical data loss.
Outcome: profiles, messaging history, campaign metrics, and user-level data access restored.
SourceWhale · Google Sheets · Apps Script · Claude Code · Perplexity Computer
Low-code operations
Business process automation
Designed low-code automations for onboarding, intake forms, job description parsing, template creation, tracking setup, messaging, timelines, deliverables, and assignments.
Outcome: automated client onboarding and removed recurring operations-manager / assistant workload.
Typeform · Make · Google Sheets · ClickUp · Discord
Technical operations
Streaming technology automation
Evaluated, implemented, and streamlined a creator-grade streaming setup across OBS, routing, scenes, capture devices, audio paths, and repeatable live-production workflows.
Outcome: repeatable live-production workflow across scenes, audio routing, capture, and streaming setup.
OBS · Twitch · Discord · multi-monitor workflow · streaming ops
How I work
I turn unclear operations into usable systems.
- 01Discover
Understand the real business workflow, not just the tool request.
- 02Map
Find the repeated steps, data sources, edge cases, handoffs, and failure points.
- 03Build
Create the AI workflow, low-code automation, spreadsheet system, or agent that removes friction.
- 04Calibrate
Test against real scenarios, add feedback loops, and tune the system for human judgment.
- 05Scale
Document, hand off, and make the workflow repeatable for the next role, team, or project.
About me
I bring operator range to technical automation work.
I’m a full-cycle recruiter turned business owner, sourcing subject matter expert, tech startup advisor, keynote conference presenter, and self-taught AI automation builder.
Outside the work, I’m a sponsored pickleball player, pastor-in-training in seminary, indoor cycling instructor, actor, master scuba diver in progress, competitive Magic: The Gathering player, and girl-dad.
Contact