Operations · Project Management ·
Business Analysis · Data
· Bangkok

Arriya Naoprakhon

Portrait of Arriya Naoprakhon
฿1M+(every month)est. recurring profit from a pricing tool scaled nationwide. Not a one-off.
24 plantsacross 11 regions on one daily price-recommendation process
150 staffacross 20 farms brought under one SOP governance loop

Two years in CPF's leadership development program, running process improvement and governance projects across 20+ farms and 24 plants, from frontline staff to top management.

Most problems show only their tip

I look below the surface to find the real cause, then build systems people actually follow.

Dive 1Just below the surface

I built this site myself with AI tools. Its design is based on the Iceberg Model from systems thinking: look below the surface to find the real cause.

A few things about me.

What I believe in, and how it shows up in my work.

Tap a bubble to pop it ↓
Each bubble holds one belief. Pop one.

Dive 2How I think

Ask why until the answer changes the fix.

A real case. Press "Ask why" to dig one level deeper each time.

Case · Farm operations, 2025

Nitrite kept spiking in the ponds.

High nitrite puts a whole crop at risk. The farm kept applying the usual fix, but it wasn't working.

  1. PROBLEMAll 4 ponds over the safe limit for weeks, despite the usual fix.
  2. ···6 more levels below. Press "Ask why".
Step 1 of 5

Dive 3The work behind the numbers

Problems, what I did, and what changed.

Swipe or use the arrows. Open any card for the full story.

Pricing · 202611 regions

Scaling a price-recommendation tool from one region to the whole country

80–90% of recommendations adopted · ฿1M+/month est. profit

before
1 region
after
11 regions
Read the full storyClose→
Problem
A daily pricing tool worked in one region, but scaling it meant handling 50–60K rows of order data every day.
Action
Automated the data flow with Excel Power Query and rolled it out to 11 regions and 24 plants. Added plant-level fill-rate views to tell true supply shortages (raise price) from single-plant gaps (send to inter-plant transfer).
Result
80–90% of daily recommendations adopted, contributing an estimated ฿1M+ incremental profit per month.
Leadership · 2026Co-sponsor · 6 trainees

Coaching 6 management trainees to plan from the target and fix problems at the root

Kept a key account and protected ~฿1M in sales

target
฿5M
plan
฿2M

Example of the rule I taught: if the initiatives don't add up to the target, it isn't a plan yet.

Read the full storyClose→
Starting point
Field work moves fast, so the natural focus is solving today's issues quickly. The opportunity was to add one more step: making sure the same issue doesn't come back, and linking daily work to the sales target.
Action
I coached them to find root causes and to plan from the target: break it into initiatives with numbers, and check they close the gap. I gave guidance and contacts, not answers, and let them do the work.
Result
A major pork customer threatened to switch after its main plant closed and deliveries from the backup plant ran late. I brought the trainees together with the logistics and plant heads to find the cause and agree a fix. The problem didn't come back, the customer stayed, and about ฿1M in sales was protected.
SOP governance · 2025150 staff

Getting 150 field staff to actually follow SOPs

Staff skipping the exam cycle: ~40% → under 5 people

↓ lower is better
before
~40%
after
<5 / 150
Read the full storyClose→
Problem
SOPs existed but weren't followed. Reported data said zero mortality while actual survival was about 50%.
Action
Revised SOPs with experts and built a closed loop: exam, retraining, retest, plus weekly learning sessions. The first accountability policy was rejected; I came back with the data gap above and got HR and top management aligned.
Result
Staff not completing the cycle fell from ~40% to under 5 of 150. Separately, 80% of sites improved biosecurity audit scores within 3 months.
Risk tracking · 2025AppSheet

A risk tracker that made problems get fixed within a week

Risks still open after a week: 40% → 10%

↓ lower is better
before
40%
after
10%
Read the full storyClose→
Problem
Field and lab teams flagged risks in chats and spreadsheets. Many were never closed.
Action
Built an AppSheet tracker shared by field and lab teams across 5 sites, with a weekly follow-up rhythm. Each week I also pulled growth, FCR and biomass data and flagged underperforming ponds to farm teams.
Result
Risks unresolved after a week dropped from 40% to 10%.
Harvest planning · 2025Team project · adoption on my farms

Turning an SOP formula into a tool farms used before every harvest

฿2–3M/year saved across 3 farms · later rolled out nationwide

Field data→SOP formulas→When & how much to harvest
Read the full storyClose→
Problem
Each crop is harvested in 3–4 partial rounds. Misjudge stock count or size and ponds get crowded, growth slows, sizes miss spec and prices drop, with knock-on effects for plants and sales forecasts. The SOP had the right formulas, but decisions were usually made from experience.
Action
My teammates had built the SOP formulas into an Excel template. The next step was getting farms to use it. Each of us took the farms we were responsible for. On mine, I introduced it to the farm managers, trained them, worked through the numbers with them every week, and made sure it was used before every harvest.
Result
About ฿2–3M saved a year across 3 farms, then rolled out to farms nationwide.
B2B sales · 2024Territory

Growing a sales territory 2.5× in six months

Market share 60% → 75%, beating target

before
฿1.2M
after
฿3.0M

Monthly sales in territory

Read the full storyClose→
Problem
Quotes started from floor price, margins were thin, and key accounts were at risk over late deliveries.
Action
Segmented customers by preferred size and buying behavior. Quoted at list price, with end-of-day discounts only to clear stock. Re-sequenced routes with logistics and checked key-account orders daily.
Result
Sales from ~฿1.2M to ~฿3M a month, +฿0.2M/month margin with no volume lost, and ฿0.3M/month in key accounts retained.

See full work history ↓

Dive 4The roots

Four rotations, one thread: making operations work better.

Charoen Pokphand Foods (CPF), Jul 2024 – Apr 2026

  1. 2026

    Art Teacher (part-time)

    Private art school · kids aged 3–10

    One-on-one classes after leaving CPF. Mostly practice in reading people and adapting how I communicate to each person.

  2. Feb 2026 – Apr 20263 months

    Business Development, Pricing & Commercial Ops

    CPFGS · Potential Leaders Development Program

    National pricing tool, plant-level fill-rate analysis, and co-sponsoring 6 management trainees.

  3. Oct 2025 – Jan 20264 months

    Business Development, Farm Strategy

    CPFTH · 20 sites

    SOP governance redesign, data checks on field reporting, and a biosecurity program.

  4. Jan 2025 – Sep 20259 months

    Project Manager, Farm Enhancement

    CPFTH · 5 sites

    10+ end-to-end improvement projects on 2–3 week cycles, 2–3 running in parallel.

  5. Jul 2024 – Dec 20246 months

    Management Trainee, Fresh Products Sales (B2B)

    CPFGS · Future Leader Program

    Territory sales, customer segmentation and pricing discipline.

  6. Education

    B.A. Agricultural Resources Administration

    Chulalongkorn University

    First Class Honors · GPAX 3.81

Dive 5Still digging

Merchant Health & Risk Monitor

A self-directed project built end to end on simulated data, to answer one question: which merchants are at risk of churning, where are they, and why is their health declining?

Source

Google Sheets

Transform

BigQuery · SQL

Report

Looker Studio
Live demo · simulated data
Merchant Health and Risk Monitor dashboard: 40 merchants split into green, yellow and red, at-risk trend by month, at-risk merchants by area and type, health score by component, and a priority list of the lowest-scoring merchantsClick to open the live version ↗

Dive 6Things I build for fun

I learn new tools by building something with them.

Small experiments, mostly with AI.

AI agents · 2026 · in progress

A world of AI agent teams

I designed a small team of AI agents, each with one clear role. Work moves between them on a shared daily schedule, and nothing goes ahead without my approval. It runs as a visual world, so I can see who is working on what.

PlannerWriterReviewerEditorApproval step

What it taught me: designing roles, handoffs and review steps. The same questions I asked about any operations process.

A cartoon map showing a team of AI agents: planner, scriptwriter, script checker and editor, each with a status bubble, and an approval inbox
The team, live: who is working, who is idle, and who is waiting for my approval.

AI video · 1-week test

Learning AI video by shipping 20 shorts

I'd never made AI-generated video before, so I gave myself a week: 20 short videos made entirely with AI tools, published to see how the platform's algorithm responds.

20videos in 1 week
~5–6Ktotal views

What came next: I turned what I learned into the content team of AI agents above, with clear roles and review steps before anything goes out.

Just for you

Leave a doodle

Drag to draw

BottomLet's talk

Open to operations, project and business analyst roles in Bangkok.

Email

arriya.naoprakhon@gmail.com