Hi, I'm Stephen Ta 👋
Operations Leader
I am an Operations Leader with a track record of scaling teams across Financial Services, Fintech, Supply Chain, and SaaS. I build autonomous systems and scalable infrastructure that allow teams to execute faster, smarter, and with greater leverage.
Experience (16 Yrs)
Meta
Shopify
Stripe
Rightsline
Bird
Federal Reserve
Operational Toolkit
Business Mechanics
I studied Economics at UMass Amherst, which sparked an obsession with learning exactly how things work. I love figuring out how goods are priced, what makes a product superior, and how companies actually make money. In operations, you have to deeply understand how your team fits into the company's broader objectives. Building the systems to hit those goals is the fun part.
High-Leverage Execution
I thrive in environments where people believe they can achieve the impossible, step by step. I am a strategic leader, a forever learner, and a jack-of-all-trades who cuts through the noise. I firmly believe that LLMs and AI workflows give us the superpower to multiply our capabilities exponentially.
Autonomous Execution
My core strength is resourcefulness. I am a data person at heart, and I don't need step-by-step instructions to get moving. Give me a complex, undefined problem, and I will dig deep, figure out the mechanics, and build the solution right alongside my team of AI agents.
Life & Hobbies
I have way too many interests to list, but my current obsessions include the NBA, watching boxing, and collecting mechanical watches. Above all else, my absolute favorite thing to do is travel the world with my family.
AI Developer Workspace (myclaw)
Workspace InfraArchitected and optimized myclaw, an autonomous terminal-based AI workspace structured around the PARA (Projects, Areas, Resources, Archives) system to maintain deep context memory. The environment coordinates script-driven agents to automate calendar tasks, draft action-oriented briefs from meeting transcripts at the start of each day, and execute commands. The workspace features custom shell hooks integrated with Meta's continuous integration systems to monitor remote build progress and landing statuses for active pull requests (PRs) directly through Mercurial stacked diff CLI outputs.
AI-Powered Capacity Planning
Workforce AnalyticsEngineered a secure capacity planning pipeline to model workforce task categories and measure operational efficiency across departments. The ingestion engine translates raw manager call notes and AI initiative reports into structured data. By validating transcript metadata using strict Pydantic schemas, the pipeline maps worker activities to specific workstreams to isolate efficiency gains and identify potential headcount (HC) changes. Results are synchronized programmatically across five dedicated logging sheets (Conversation Logs, Capacity Impact, Skills Gap, Action Items, and Engagement Tracker) to guide leadership on resource reallocation.
Shopify Services Org
Architected a net-new Professional Services business unit from scratch. Transformed a legacy "free work" culture into a structured operational and billing model.
Defined global service packages, established staffing and recovery KPIs, integrated services into pre-sale workflows, and aligned compensation structures to achieve a self-sustaining break-even financial model.
Resource Forecast
Built ML-driven allocation models utilizing 18 months of historical PSA and Salesforce data. Tested Scikit-Learn Random Forests against Linear Regression models.
Deployed a human-in-the-loop validation framework where program managers graded accuracy to iteratively weight algorithm coefficients. Integrated insights directly into pricing structures.
Vendor Spatial Capacity Platform
Designed a vendor spatial capacity platform to monitor and manage physical desk allocation across offices. The pipeline uses an automated n8n workflow to aggregate vendor-provided template reports into a structured database, presenting unified dashboard tables categorized by city, building, floor, business unit supported, and shift headcount compared against maximum capacity.
The engine compares current active headcount against Anaplan plans and physical facility capacity limits to determine scale readiness. It runs bi-weekly AI insight sweeps to flag capacity risks and uses an AI-powered booking helper to suggest optimal floor allocations. Bookings submitted through the AI side panel trigger interactive webhooks routing to Google Chat bots for approval, automatically firing email notifications to Anaplan team members to update the staffing plan.
Investment Fund Review Platform
Transformed a static investment fund business review deck into a light interactive web application and automated email summaries. Using Python scripts and dynamic HTML templates built via custom Claude Code skills and integrated with n8n workflows depending on the use case, the system aggregates budget data and metrics. Bypassing manual report screenshots, it compiles and distributes high-fidelity, responsive CSS waterfall charts and executive text briefs directly to leadership.
Live Waterfall Visualizer
Variance: -19.4%
Technical Code Sandbox
BI Dashboard-to-Slides Automation
A Python automation pipeline that fetches raw data from the in-house BI Dashboard API, renders pixel-perfect HTML/CSS charts, captures them using a headless browser, and generates Google Slides decks via CLI.
import subprocess
from pathlib import Path
def generate_slides_from_metrics(widget_id: str, deck_id: str = None) -> str:
"""Fetches dashboard metrics, renders an HTML chart, takes a screenshot, and uploads to Slides."""
# 1. Fetch metrics from the BI Dashboard (allow_uii = True)
data = fetch_dashboard_metrics(widget_id, allow_uii=True)
# 2. Render pixel-perfect horizontal bar chart in local HTML
max_val = max(item['value'] for item in data['series'])
bars = "".join(f"""
<div style="display:flex; margin:10px 0;">
<div style="width:120px;">{item['label']}</div>
<div style="flex-grow:1; background:#eee; height:20px; border-radius:4px; margin:0 10px;">
<div style="width:{(item['value']/max_val)*100}%; background:#6EADFF; height:100%;"></div>
</div>
<div>{item['value']} FTE</div>
</div>""" for item in data['series'])
html_content = f"<html><body><h2>{data['title']}</h2>{bars}</body></html>"
Path("temp.html").write_text(html_content, encoding="utf-8")
# 3. Headless browser screenshot
subprocess.run(["browser_screenshot", "temp.html", "chart.png"], check=True)
# 4. Insert image into Google Slides via enterprise CLI
cmd = ["gslides", "add-image", "--image", "chart.png", "--title", data['title']]
if deck_id: cmd.extend(["--deck-id", deck_id])
result = subprocess.run(cmd, capture_output=True, text=True, check=True)
# Clean up temp files
Path("temp.html").unlink(missing_ok=True)
Path("chart.png").unlink(missing_ok=True)
return result.stdout.strip()Local AI & Personal Infrastructure
I am currently exploring the limits of local, privacy-first AI using my own hardware. I'm building automated IoT workflows around my house, alongside self-hosted personal finance and family planning tools. Beyond building, I am actively running training sessions to teach other operators and teams how to practically leverage AI to multiply their own impact.