pipesline
Headshot of Merek Pipes in a navy suit

$ cat ~/README.md

Merek Pipes

I love to solve problems in efficient ways that alleviate the burden of thought from the user.

I grew up on a cider farm watching a family business grow, played goalie across Canada and the US, studied Computer Engineering with a minor in Economics, and now I build AI systems that run in production.

studied
Computer Engineering, Economics minor
gpa
3.97, graduated early
hockey
goaltender, 4 years NCAA Division I
now
AI systems · Azure · full-stack
starter
sourdough, free on request
see my work hello@pipesline.com

$ cat roots/merridale.md

Growing up in a growing business

My parents bought Merridale Ciderworks when I was 2 years old, and I grew up on the farm watching it grow from just 2 part-time employees to the 40+ full-time employee corporation that it is today.

2part-time employees when we started
40+full-time employees today

Throughout that time, I've helped modernize a number of their systems and technical solutions. From inventory, to their sales pipeline, and beyond.

  • inventory
  • sales pipeline
  • data visualization
  • …and beyond
Growing up watching a family business grow has given me insight into how small businesses run: what works and what doesn't, and what's possible (and what's not) for people who don't come from a technical background.

$ traceroute union-college

Hockey took me across the map

I grew up in Cobble Hill, BC, and played hockey all across Canada and the United States. It took me from Shawnigan Lake, to Richmond, to La Ronge, Saskatchewan, to Swan Valley, Manitoba, and all the way to Schenectady, New York, where I got a hockey scholarship to play goal for Union College.

  1. Shawnigan LakeBritish Columbia
  2. RichmondBritish Columbia
  3. La RongeSaskatchewan
  4. Swan ValleyManitoba
  5. SchenectadyNew York · Union College
4years at Union College, NCAA Division I
4×Thomas Van Arden Dukehart Academic Award, highest GPA on the team

$ cat academic/transcript

Union College

A degree in Computer Engineering with a minor in Economics. I graduated early with a 3.97 GPA while juggling hockey, school, and research.

3.97GPA
earlygraduated ahead of schedule

research/

For my 3rd and 4th year, I did research with Professor Chandra Pappu in using chaos theory for a joint radar and communication waveform: controlled chaotic signals for dual-functional bistatic radar-communication systems. The idea is to integrate radar with communications so you need fewer RF transmitters.

two Lorenz signals that start 0.000001 apart. deterministic, but they look like noise.

awards/

  • IEEE AESS Engineering ScholarshipGiven to only 2 people annually across the country, for research in the aerospace field. The first recipient from Union College.
  • Top Computer / Electrical EngineerAwarded to the most excellent Computer or Electrical Engineer in my year.
  • 4× AHCA All-American ScholarAmerican Hockey Coaches Association.
  • ECAC Hockey All-Academic Team
  • 4× Thomas Van Arden Dukehart Academic AwardHighest GPA on the Union men's hockey team.

$ ls work/

Two systems, both running in production

At work, I've built two different systems. Both are full IaC setups in Azure.

01 · graph rag · prediction

Time entries that write themselves

It pulls data from a number of integrations and stores it on a synced schedule. It then makes predictions on that data using a graph RAG system to create time entries for billable and non-billable hours against jobs.

  • GitHub
  • Google Calendar
  • Google Drive
  • Outlook Calendar
  • OneDrive
  • Jira
  • Zoom
  • etc.

new time entryexample

job
task
hours
billable
notes

A time entry has a lot of fields (job, task, hours, and more) that it needs to accurately guess, so it uses both heuristics and classification. It falls through a tiered system, first to last:

  1. history for the client learned behaviour
  2. history of teammates
  3. full-text search
  4. vector search
  5. LLMs
  6. …and more

02 · agent · reporting

A chatbot that can actually do things

A chatbot that has sandboxed access to do reporting, as well as take actions like approving timesheets. It can create charts for a user, tracks the history of how people are using it, and has proper token tooling for security.

  • sandboxed reporting
  • approves timesheets
  • builds charts
  • usage history
  • token security

also/

Outside of that, I keep up to date on all the models, and I have extensive experience in how best to use AI within a corporate environment in the most secure way possible. Before this I was doing a lot of full-stack web development, and I have a ton of experience on the web side of things.

  • Azure
  • IaC
  • Graph RAG
  • vector search
  • LLM agents
  • secure enterprise AI
  • full-stack web

$ ls built/

Things I've built

operations · live ↗ qclstaffboats.com Staff scheduling and boat booking for a fishing lodge. product · live ↗ rentive.ca Rental marketplace with AI support and AI listing import. Built solo.
data viz Merridale sales pipeline Data visualization of the sales pipeline at Merridale Ciderworks.
hardware · ml Remote weather station A remote weather station powered by machine learning.

$ cat offline/*.txt

When I'm not coding

  • dog sitting
  • working out
  • BJJ
  • cooking + baking

I would happily share sourdough starter with anyone who would ask.

status: hungry

$ ./contact

Want to work together?

hello@pipesline.com what I offer → github ↗