Open to founding ops and Chief of Staff roles

Toronto ET

Max WangFounding-team operator. I build the systems early companies run on.

Ops, logistics, finance, and AI tooling, built from zero and shipped into live operations. Based in Toronto, open to relocating to the US.

Email me LinkedIn (opens in a new tab)Resume PDF

The ledger

Figures from shipped work. Each links to its file.

  1. 32,000 → 10,000leads qualified by a Python pipelineFile 01
  2. 270+ hrsof manual work savedFile 01
  3. 6 wks → 1investor list enrichment time, after automationFile 01
  4. $2Mraise the investor outreach helped closeFile 01
  5. Hours → minsinbound lead response time, with an AI draft ready almost instantlyFile 02
  6. −60%East Coast and Midwest shipping costsFile 03
  7. 50+ → 12SKUs after the catalog overhaulFile 03
  8. −20 mlfill per can, adding headroom that drastically cut burstsFile 04
  9. ~$0.10lower COGS on every canFile 04

Selected work

Same format every time: the problem, what I built, and what changed. Open any file for the full write-up and a diagram of how it works.

Investor Pipeline That Helped Close $2M

A Python pipeline that turned a raw lead list into an investor CRM and automated outreach for the fundraise.32K leads → 10K qualified, 270+ hours saved

Context Shift Naturals, hemp beverage startup. Chief of Staff.

Problem

The fundraise needed a targeted investor list, and manually sorting a raw list of 32,000 leads would have taken hundreds of hours.

What I built

A Python pipeline that filtered and enriched the raw list down to 10,000+ angel, venture, and family office contacts. Internal automation replaced the manual verification step.

That list became the company’s investor CRM, with automated outreach workflows built in Google Apps Script.

Result

Saved 270+ hours of manual work and cut enrichment time from 6 weeks to 1 week. Outreach ran at a ~1% reply rate and ~0.3% meeting conversion, and fed the raise that helped Shift close $2M.

RAW LEAD LIST32,000100%PYTHON PIPELINEfilter, enrich, verify270+ HRS SAVED6 weeks down to 1ANGEL, VC + FAMILY OFFICE CONTACTS10,000+~22,000filtered out~31% keptCRM + AUTOMATED OUTREACHApps Script · ~1% reply rateHELPED CLOSE A $2M RAISE
Fig. 01Lead pipeline. Bar widths are to scale.

Inbound AI for a Real Estate Developer

An AI that drafts the first reply to every inbound lead, grounded in the company's own website and shaped by the team's edits.Lead response time cut from hours to a few minutes

Context Reurbano, Mexico City real estate developer. Special Projects.

Problem

Inbound leads needed fast, accurate, on-brand replies, and writing each one from scratch slowed the team down.

What I built

An Inbound AI that drafts the first response to every incoming lead and pings team members the moment the draft is ready. It grounds its answers by crawling Reurbano’s own website, and it learns from the team’s edits, so drafts get closer to how the team actually writes over time.

A deterministic safety layer sits outside the learning loop, so the rules on what it can and can’t say never drift as it learns.

Also designed cohort segmentation logic that automatically sorts each inbound lead into the right group, so follow-up is targeted.

Result

A draft is ready almost instantly and the team is pinged right away, so lead response times dropped from hours to a few minutes. The team edits instead of writing from scratch, and response quality improves without anyone retraining it.

INBOUND LEADSITE KNOWLEDGEcrawled from thecompany websiteLEARNED STYLEshaped by theteam's own editsDRAFTS FIRST REPLYSAFETY LAYERfixed rules, outside the learning loopteam pinged instantlyTEAM EDITS + SENDSevery edit feeds the learned styleLEAD RESPONSE TIMEHOURS TO MINUTES
Fig. 02The safety layer sits outside the learning loop, so the rules never drift.

Two-Warehouse Fulfillment and Catalog Overhaul

Shopify order routing across two 3PLs that cut East Coast and Midwest shipping costs by 60%, plus a catalog cut from 50+ SKUs to 12.60% lower East Coast and Midwest shipping costs, 50+ SKUs cut to 12

Context Shift Naturals.

Problem

Inventory was split across two warehouses with different stock, and the catalog had grown to 50+ SKUs.

What I built

A two-warehouse Shopify fulfillment routing system from scratch across Collective Pallet (Oregon) and Grayland (Tennessee), including shipping profiles and split-order logic. Tennessee held only the cocktails; Oregon held the full sparkling water line.

Cut the catalog from 50+ SKUs to 12. Also built the Shopify theme with custom Liquid, CSS, and JS, configured tiered subscription discounts, and built an inventory tracker in Google Sheets with Apps Script controls.

Result

Shipping costs to the East Coast and Midwest dropped by 60%. Orders routed to the right warehouse automatically, and the catalog became much simpler to manage and sell.

SHOPIFY ORDERPRODUCT CHECKCOCKTAILSGRAYLANDTennesseeMIXED CARTSPLITships from bothSPARKLINGCOLLECTIVEPALLETOregonSHIPPING COST, EAST COAST + MIDWESTTO SCALEBEFORE100%−60%after routingCATALOGTO SCALE50+ SKUSbefore12 SKUSafter the overhaul
Fig. 03Every order is checked against what each warehouse actually holds.

Fixing Exploding Cans With a Fill-Level Study

A fill and headroom study that took 20 ml out of every can, giving it room to breathe and drastically cutting bursts in transit.~$0.10 lower COGS per can, drastically fewer bursts in transit

Context Shift Naturals.

Problem

Cans were filled to the very top with no headroom, and they were bursting in transit. The fill level was also driving up cost.

What I built

A fill and headroom case study benchmarked against category peers, which led to reducing fill by 20 ml per can to create headroom.

Result

Drastically reduced cans bursting in transit, and lowered COGS by about $0.10 per can.

BEFOREfilled to the very top0 ml headroomHEADROOM20 mlAFTER20 ml less fillroom at the topBURSTS IN TRANSITDRASTICALLY REDUCEDCOGS PER CAN~$0.10 LOWER
Fig. 04Illustrative, not to scale. Before: filled to the very top, no headroom. After: 20 ml less fill leaves headroom in every can.

Regulatory Compliance for a 22,464-Can Relabel

OLCC hemp registration and label compliance, with a 22,464-can relabel prepared across the canning and label partners.Compliant labels ready for 22,464 cans

Context Shift Naturals.

Problem

Products had to meet Oregon (OLCC) hemp labeling rules, and the nutrition panel needed to change from 1-serving to 2-serving.

What I built

Owned OLCC hemp product registration and label compliance (OAR 845-026-6050, ASTM D8441 symbol requirements, Label ID, COA QR codes). Prepared the relabel of 22,464 cans with the canning and label partners.

Result

Compliant labels were ready for the full run of 22,464 cans. The relabel itself never went ahead, because Shift began winding down operations.

Label specStatus
  • OLCCHemp product registration
  • OAR 845-026-6050Hemp label compliance
  • ASTM D8441Symbol requirements
  • Label ID
  • COA QR codesCertificate of analysis
  • Nutrition panel1 serving → 2 servings
  • Relabel runShelved when Shift began winding downShelved
22,464cans, compliant labels ready
Fig. 05The compliance work was finished; the relabel itself was shelved in the wind-down.

Pricing the Wind-Down of ~86,000 Cans

Re-engaged for a time-sensitive wind-down to build the liquidation and pricing analyses behind the recovery decisions.Liquidation and pricing scenarios across ~86,000 cans

Context Shift Naturals. Special Projects, re-engaged for the inventory liquidation and wind-down.

Problem

Shift was winding down and had to liquidate its remaining inventory on a tight timeline, with founders, investors, warehouse partners, and prospective buyers all involved.

What I built

Liquidation and pricing analyses across approximately 86,000 cans, built on COGS, freight, channel economics, and bulk-sale scenarios.

Coordinated the founders, investors, warehouse partners, and prospective buyers around those numbers.

Result

The analyses support the recovery decisions on the remaining inventory. The work is ongoing.

INPUTSCOGSFREIGHTCHANNEL ECONOMICSBULK-SALE SCENARIOSLIQUIDATION + PRICING MODEL~86,000 cansRECOVERY DECISIONSCOORDINATED WITHfounders · investorswarehouse partners · prospective buyers
Fig. 06Every scenario runs on the same inputs, so founders and investors compare like with like.

Investor Pre-Brief Agent

An event-driven agent that researches an investor with AI and data enrichment, then posts a pre-brief to Slack.Built end to end: CRM event in, researched brief in Slack

Context Reurbano, Mexico City real estate developer. Special Projects.

Problem

Preparing for investor conversations meant manual research on each person and firm.

What I built

An event-driven agent. An Attio webhook triggers a service on Railway, which has Claude research the investor and enrich the profile using web search and several databases. A Slack bot then posts the pre-brief to the team.

Also built a WhatsApp-to-Attio workflow that routes inbound leads straight into the CRM.

Result

Built and working end to end, from the Attio webhook to a brief posted in Slack. It was never put into official use.

01ATTIO WEBHOOKCRM event fires02RAILWAY SERVICEcatches the event03CLAUDE + ENRICHMENTweb search + databases04SLACK BOTposts the pre-briefALSO BUILTWHATSAPPinbound leadATTIO CRM
Fig. 07Event-driven: nobody has to remember to run it.

Co-Founded an Ecommerce Brand

Built the operational backbone of a brand from scratch, then taught myself video to cover a marketing gap.Built the operational backbone from scratch

Context Synapse, an ecommerce brand I co-founded (now exited).

Problem

A new brand with no operational backbone, and a marketing gap nobody was covering.

What I built

Much of the company’s operational backbone, from scratch: ecommerce, fulfillment, inventory, supplier coordination, customer operations, pricing, and internal systems. Owned the problems that didn’t yet have an owner.

Taught myself AI video generation and editing in Adobe Premiere to cover the marketing gap.

Result

Manual processes turned into scalable workflows, and those videos drove sales. Core systems were documented and handed over during the partner transition.

How I work

  1. Zero to one.

    I'm happiest building the first version of a system where none exists yet.

  2. Own the unowned.

    I pick up the problems that fall between roles and get them closed.

  3. AI as leverage.

    I use AI to ship tools and workflows that would normally need a dedicated engineer.

  4. Details matter.

    Compliance specs, reconciliations, fill levels: small numbers add up to real money.

Background

Experience

  • Shift NaturalsHemp beverage startup
    • Special ProjectsSep 2026 – Present
    • Chief of StaffAug 2025 – Jul 2026
    • Bizops AssociateJan 2025 – Aug 2025
  • SynapseEcommerce · Toronto
    • Co-Founder, Operations & StrategyJun 2026 – Sep 2026
  • ReurbanoMexico City · Remote
    • Special ProjectsMay 2026 – Jun 2026
  • Insight Academy of CanadaToronto
    • Chief of StaffJan 2024 – Dec 2024
  • GAC China
    • Operations InternJul 2023 – Dec 2023

Education

University of Toronto

B.A. Statistics and Economics

Dec 2023

Also

Fluent in Mandarin.

Family background in pharmaceuticals.

Looking for my next founding ops or Chief of Staff role.The fastest way to reach me is email.

Email me LinkedIn (opens in a new tab)Resume PDF