sergey.desyatkov
YVR

~/$ whoami

Sergey Desyatkov

Senior Software Engineer / Autodesk / Vancouver, BC

I build platforms
— the quiet systems products stand on.

years shipping
10+
product companies
5
clouds in prod
AWS·GCP·Azure
AI tooling
agents · MCP
tree ~/career --grow · springclick a blossom to open it

01

Ask me anything

$ sergey --interactive

type or tap a command — windows open on the right.

sergey@yvr: ~zsh

02

Selected work

$ ls ~/work --sort=impact

windows are draggable — rearrange the desk.

ai-code-review.platform2026

Autodesk

An AI code-review platform, designed and shipped

Took an internal AI code-review platform from design to production. Lead reviewer and de-facto support owner — the person teams page when the bot has opinions.

  • LLMs
  • Claude · Bedrock
  • Python
  • TypeScript
  • Node.js
workflow-engine.kafkaSimpplr
┌─ step ─┐    ┌─ step ─┐
│   A    │───>│   B    │──┐
└───┬────┘    └────────┘  │
    │         ┌─ step ─┐  ├──> done
    └────────>│   C    │──┘
              └────────┘

A workflow engine, from scratch

Built a Kafka-based workflow manager that wires many company services into fully managed flows — parallel and sequential steps, with end-to-end monitoring. The most hardcore thing I've engineered.

  • Kafka
  • NestJS
  • event-driven
membership.rebuildAMA

Alberta Motor Association

A membership platform, rebuilt end to end

Front to back — including the React tool agents use to service member accounts. Moved SQL Server to Aurora PostgreSQL with DMS CDC and Glue/PySpark ETL, and validated integrity across 1M+ records.

  • AWS CDK
  • Prisma
  • React
  • PySpark
ledger.rewardsSimpplr

An auditable rewards ledger

Led a ledger-based rewards platform and a sync layer handling millions of daily events across platforms, with end-to-end tracing through the microservices.

txaccountΔ pts
0x3e1acct_7f…+250
0x3e2acct_7f…−100
0x3e3acct_a2…+40
  • Kafka
  • PostgreSQL
  • Terraform
  • K8s
idp.backstagenow

The platform engineers live in

Maintaining Autodesk's Internal Developer Platform on Backstage. Building backend services for AI-agent automation — ADRs, MCP integrations. Also the team's scrum master.

  • Backstage
  • MCP
  • React
  • PostgreSQL
monolith→servicesWalkMe

Splitting the monolith

Helped move a monolith to microservices across AWS, GCP and Azure — GraphQL, gRPC and REST APIs, CI/CD on Docker and Kubernetes.

migrate.pywip

side project · private

A config-driven migration tool

Database-agnostic migrations where one config describes type conversions, transformations and validation. Born from too many SQL Server → Postgres nights.

  • Python
  • PySpark
  • Temporal

03

A literal file tree

$ tree ~/career --grow

hover a folder to find its branch.

~/career/

current branch   grown  ·  spring

04

Commit history

$ git log --since=2014 --graph

one pixel per month. hover to read it.

  1. a1f3e9c2026feat(autodesk): ship an AI code-review platform
  2. 7c21b042025refactor(ama): rebuild membership platform end to end
  3. 3e88d212024feat(simpplr): kafka workflow engine + ledger rewards
  4. 91ab44f2022refactor(walkme): monolith → microservices, three clouds
  5. 5d02cf12016init: first commit as a frontend developer
  6. 0f15b7e2015chore: complete full-stack program, John Bryce
  7. 00000002003art: diploma in fine-arts education

05

From fine art
to fine-grained systems

I trained as a fine-arts teacher. Then I found out code is also a medium.

My first diploma (Magnitogorsk, 2003) is in fine-arts education — later assessed as bachelor's + master's equivalent. In 2015 I went through an intensive full-stack program, and a year later shipped my first production frontend.

Ten years on I mostly live in the backend: event pipelines, ledgers, migrations, platforms. The artist never left — he just draws boxes and arrows now, and cares a lot about how the tool feels to the person using it.

Off the clock: chess, cooking, games — and pixel art, obviously.

stack.config.tsTS

            
          

full cycle

From idea to production, and back again.

I don't stop at tickets. I work the whole loop: shaping the problem with product, designing and architecting it, building front to back, shipping it, watching it run, and helping the team get better along the way.

01 / 07

Product

    $ echo "what's next?"

    Have something worth building? Let's talk.

    LinkedIn

    Vancouver, BC · Pacific time