You share your idea.
We get it done.

Code Reflect is a senior engineering studio in Belgrade. We design, build and run production software: web, mobile, backend, and the AI features on top. For founders with an idea, and for teams with more roadmap than engineers.

A few clients at a time Belgrade · building since 2021 Go · PostgreSQL · TypeScript Your repo, your cloud, your IP
GoPostgreSQLpgvectorTypeScriptSwift / iOSCloud RunKubernetesNATSTerraformOpenTelemetryAnthropicOpenAIRedisgRPCKafkaPython

Getting software built usually disappoints.

If you cannot build it yourself, you have three well-worn options. Each one leaks in its own way, usually after the contract is signed.

The big agency

Quotes fast, staffs slow. The senior people from the sales call are gone by the kickoff, the juniors who replace them learn on your budget, and the invoice never notices the difference.

The lone freelancer

Brilliant until life happens. One brain holds the whole system, nobody reviews the code, and when they move on, the knowledge moves with them. Bus factor: one.

Hiring in-house

Right answer long-term, wrong answer this quarter. A good team takes two quarters to assemble and a bad hire costs a year. Your roadmap does not wait that long.

We built Code Reflect to be the fourth option: a small senior team with skin in the game.

Build. Extend. Rescue.

Three jobs we get hired for. Different starting points, same standard: software that runs in production and a team that answers for it.

Build your product from zero

From a blank repo to paying users: discovery, architecture, design and delivery in one team. Web, mobile and backend, including the AI features your product actually needs, built as engineering rather than magic.

Go · TypeScriptPostgreSQLiOS · webLLM featuresGCP · AWS

Extend your team

Senior engineers inside your repo, your standups and your review culture. They raise the bar while they ship: better tests, cleaner boundaries, and juniors who level up by working next to them.

embedded seniorscode reviewmentoringmonthly rolling

Modernize and rescue

Stalled projects, abandoned codebases, systems nobody dares to touch. We audit what exists, stabilize what runs, and migrate the rest in slices, without a big-bang rewrite that bets the company.

auditsincremental migrationobservabilityCI/CDzero-downtime cutover

Four steps, in this order, on purpose.

Feasibility before scope, one real path before ten half-paths, measurement before feature breadth. It is slower for two weeks and faster for the following year.

STEP 01

Week zero

3–5 days

We map the use case, the data, the realistic failure modes and the budget. You get an architecture blueprint, a risk list and a cost-of-ownership estimate, yours to keep, whoever ends up building it.

STEP 02

Thin slice

weeks 1–3

One complete path, end to end, in your cloud, behind a feature flag. Not a demo on a laptop, but a deployable that a real user could hit, with the auth and data model already right.

STEP 03

Measure, then widen

ongoing

Tests, tracing and cost visibility go in before feature breadth. Once you can see quality drop, adding surface area stops being a gamble and starts being a decision.

STEP 04

Hand over

from day one

Documentation, runbooks and pairing with your engineers throughout, not a knowledge dump at the end. We stay while it is useful and leave a team that does not need us.

Three ways to start.

Pick by how much certainty you have. Less certainty, smaller commitment: a scoped spike answers the question before anyone signs up for a roadmap.

Scoped spike

Lowest commitment

Two to three weeks, fixed price, one question answered with working code: is this feasible, at what quality, at what cost. You keep the code and the write-up either way.

Fixed price · 2–3 weeks
Deliverable: running slice + findings

Embedded engineers

Most common

Our engineers work inside your repo, your standups and your process, reporting to your lead. Time and material, monthly rolling, scale up or down with a month's notice.

Time & material · monthly rolling
Daily stand-ups, weekly written report

Product squad

Full ownership

A self-managed cross-functional team of engineering, design and delivery that owns a roadmap and reports on outcomes instead of hours. For when you need a product built, not a seat filled.

Monthly retainer · quarterly roadmap
Design, build, run

On equity: occasionally we take part of the fee in equity: typically one startup a year, always alongside cash, and never before we have worked together for six months.

A small team that answers for its work.

We have built systems where being wrong was expensive: financial platforms, healthcare, government. That instinct, that someone has to answer for the software, is the whole culture of the studio.

Seniority

Senior engineers only

The people in the discovery call are the people writing the code. No juniors billed at senior rates, no handover to a delivery pod you have never met.

Depth

Backend depth by default

Go and PostgreSQL, distributed systems, multi-tenant architecture, financial-grade correctness. The pretty interface is the last mile; the system underneath is the product.

Ownership

Everything is yours from day one

Your repository, your cloud accounts, your IP. No proprietary wrapper, no lock-in layer, nothing that makes leaving us expensive.

Overlap

European hours, US overlap

Based in Belgrade (CET). Full overlap with Europe and the UK, an afternoon of overlap with US East. Slack, Zoom and a written weekly report.

Straight answers.

Can you build AI features into our product?
Yes, and we treat them as engineering rather than magic: model access behind an interface, regression tests on your data, cost and latency budgets from day one. We are provider-agnostic across Anthropic, OpenAI, Google and open-weight models, and we build so the choice stays reversible.
Can you take over an existing codebase?
Yes, and it is a large part of what we do. We start with a review of architecture, data model, security boundaries, and test and eval coverage, then come back with a written assessment and a prioritised plan. You can act on it with us or without us.
Do you work fixed-price?
For scoped work with a clear finish line (a spike, a migration, a defined integration), yes. For open-ended product development, fixed price forces both sides to guess badly, so we work time and material or on a retainer with a roadmap reviewed every quarter.
What does the first month look like?
Week zero is discovery and blueprint. Weeks one to three are the thin slice: one real path deployed to your environment. By the end of the month you have running software, a realistic cost picture, and a clear view of whether to widen or stop.
How do you handle NDAs, IP and data?
NDA before the first technical conversation if you want one, full IP assignment in the contract, and work in your accounts under your data-processing rules. Where regulation applies, as with health data, financial records, or the public sector, we design for it from the blueprint rather than retrofitting.
How quickly can you start?
Discovery usually within one to two weeks of the first call. Build capacity depends on what is running at the time — tell us your target date early and we will be honest about whether we can meet it.

Tell us what you are building.

Send a few paragraphs about the problem, the users and the constraint that worries you most. You will get back questions, a rough architecture and an honest price range, before anyone talks about a contract.