Independent software studio

Engineering for ambitious software.

Pxlhut builds multi-tenant SaaS, AI systems and the products that run on them. Four in the works— and we're only getting started.

Be first to try each product as it ships · No spam, ever

Building something of your own? We take on select client work →

Pxlhut Studio4 in development
Currently building 4 products — more on the roadmap
IELTSHUTIn development
IELTS prep, mock exams & band scoring
72% completeTarget Q3 2026
CrickHutBuilding
Real-time cricket scoring & fan app
48% completeTarget Q4 2026
Website BuilderAlpha
Drag-and-drop site builder (Forge)
61% completeTarget Q3 2026
Coaching PlatformDesign
All-in-one operations OS for coaching
27% completeTarget 2027
More products in the pipeline
An independent studio, building in public —
Multi-tenant SaaS
AI engineering
White-label ready
AWS · GCP native
TypeScript end-to-end
Founder-led engineering
Multi-tenant SaaS
AI engineering
White-label ready
AWS · GCP native
TypeScript end-to-end
Founder-led engineering
4 products in development
Realtime infrastructure
Edtech · Fintech · Sportstech
Roadmap-driven
Built in public
More on the way
4 products in development
Realtime infrastructure
Edtech · Fintech · Sportstech
Roadmap-driven
Built in public
More on the way
i.What we build

Four disciplines. One studio.

From multi-tenant SaaS platforms to embedded AI copilots — we ship products that survive scale, audits, and Monday morning standups.

i.

SaaS platform development

Multi-tenant architectures with white-label theming, billing, role-based auth and audit-grade compliance — engineered to scale from MVP to enterprise without rewrites.

Multi-tenancyStripe BillingRBACWhite-label
ii.

AIsystems & copilots

LLM-powered workflows, RAG pipelines and domain-tuned copilots that sit inside your product — not bolted on. Evals, guardrails and observability included as standard.

LLMRAGEvalsVector DB
iii.

Automation& orchestration

Workflow engines, event-driven pipelines and integration mesh that eliminate the manual seams between your tools. CI/CD, queues, schedulers, retries — bullet-proof by default.

TemporalKafkaWorkflowsSchedulers
iv.

Internaltools & dashboards

Operator consoles, admin panels and revenue dashboards that let your team move at the speed of your customers. Real-time, granular, and beautiful enough to actually use.

Real-timeDashboardsAdmin UIsAnalytics
ii.Our products

Four products. One platform DNA.

Four products in active development, all on the same multi-tenant chassis — so each inherits the same speed, security and white-label flexibility from day one.

Edtech · In development

IELTSHUT — an IELTS platform that grades writing.

End-to-end coaching platform with AI-graded writing & speaking, mock tests, and white-label branding for every coaching institute. Furthest along — private beta this quarter.

Writing Task 2
Band 7.5 · AI graded
72%
Build progress
Beta
Next milestone
Q3 ’26
Target launch
View full build progress
Sportstech · Building

CrickHut

Real-time cricket scoring, fantasy and analytics — built for match-day crowds.

48%
Build progress
Q4 ’26
Target launch
View build progress
No-code SaaS · Alpha

Website Builder

Drag-and-drop site builder white-labeled for agencies — editor, CMS, deploy in one.

61%
Build progress
Q3 ’26
Target launch
View build progress
LMS · Design

Coaching Platform

Multi-tenant LMS — courses, payments, attendance, parent comms, branded per campus.

27%
Build progress
2027
Target launch
View build progress
iii.The engagement, in five chapters

How a Pxlhut project unfolds.

Scroll to advance
01
Chapter 01 — Discover

Every product starts with a sharper question.

Before a single component is built, we sit with your founders, operators and customers to figure out what's actually broken — and what's already working.

Discovery sprintsUser researchTech audits
Workshop · Day 2
In session
02
Chapter 02 — Architect

We design systems that survive scale.

Multi-tenancy, event sourcing, queues, caches, idempotency — boring infrastructure decisions made now save weeks of fire-fighting later. We make them before the first commit.

System designSchema modelingCapacity planning
service API
postgres (multi-az)
redis (cluster)
queue (sqs)
worker × 8
03
Chapter 03 — Build

Weekly demos. Cinematic craft.

We ship working software every week — not Figma decks. Frontend, backend, infra and DevEx march in lockstep so you see real product, not roadmaps.

Weekly demosTrunk-basedPreview envs
Sprint 14 · Velocity
+22%
04
Chapter 04 — Launch

Production-grade from day one.

Observability, SLOs, runbooks and rollback plans before launch — not after the first incident. We ship products you can actually be on-call for.

SLOsRunbooksOn-call
Launch Day · Status
All green
API Latency42ms ✓
Error Rate0.02% ✓
DB Pool38% ✓
05
Chapter 05 — Scale

We don't disappear after launch.

Ongoing engineering, optimization, growth experiments and platform evolution — Pxlhut stays embedded as long as the product is alive. Partners, not vendors.

RetainersGrowth engPlatform care
ARR · 12 months
iv.Under the hood

A stack chosen for speed and survival.

Boringly reliable. The same primitives the world's largest engineering teams bet their roadmaps on — composed into the platform layer every Pxlhut product inherits.

Languages
TypeScript · Go · Python
Primary cloud
AWS · GCP multi-region
Security
Privacy-first · GDPR ready
Status
In development
01 / Top
Experience layer
Interfaces customers and operators actually touch.
NNext.js 16app router
RReact 19RSC
TTypeScriptstrict
TTailwindtokens
GGSAPmotion
SStorybook
02 / Middle
API & services
Type-safe RPC, idempotency, retries, rate limits — by default.
NNode · Bunruntime
GGohot paths
ttRPC · GraphQL
FFastify
TTemporalworkflows
KKafkaevents
03 / Base
Data & storage
Multi-tenant by design. Hot, cold and vector, all observable.
PPostgreSQLprimary
RRediscache · queue
MMongoDB
EElasticSearch
PPineconevector
SS3 · GCS
04 / Intelligence
AI & ML
Models, retrieval, evals and guardrails as a first-class layer.
OpenAIGPT-4o
AAnthropicClaude
LLangChain
LLlamaIndex
WWeights & Biases
RReplicate
05 / Foundation
Infrastructure & ops
IaC everywhere. Zero-downtime deploys. Observable by default.
AAWSprimary
GGCP
KKubernetes
TTerraform
GGrafanao11y
SSentry
Every product Pxlhut ships inherits this exact stack — versioned, patched and monitored as a single platform.
Request full architecture doc
v.Where we are

Honest progress, not vanity metrics.

We're pre-launch and building deliberately. Here's exactly where things stand — updated as we ship, no edits.

0
Products in active development

IELTSHUT, CrickHut, Website Builder and Coaching Platform — all built on one shared multi-tenant chassis, with more on the roadmap.

0%
Lead product progress
IELTSHUT — private beta this quarter.
Q3 ’26
First launch window
Starting with IELTSHUT and the builder.
0%
Built in-house
Founder-led engineering — no outsourcing.
Heads-down right now
IELTSHUT is heading into private beta

The exam engine, AI grading and the proctoring kiosk are already built — we're piloting with real coaching centers before the public launch.

See the full build progress
vii.AI & Automation

AI that ships, not just demos.

We embed copilots, retrieval and automation directly inside products — with the evals, guardrails and cost controls that keep them trustworthy in production. It's the same stack grading writing inside IELTSHUT today.

Production copilots

Domain-tuned assistants that live inside your product — reading your data, following your rules, and doing real work for your users.

copilot · /v2/agentrunning
OUT3 clauses need review — §4.2 liability cap, §7 auto-renewal, §9 data residency
guardrails passed · 0 PII · eval 0.97

Retrieval (RAG)

Grounded answers from your own docs and data — with citations, not hallucinations.

Agentic automation

Multi-step agents that handle support triage, ops workflows and reporting.

Evals & guardrails

Hallucination tracking, PII scrubbing and prompt-regression tests in CI.

Voice & transcription

Speech-to-text, natural TTS and pronunciation scoring — already live in IELTSHUT.

Cost & observability

Per-call budgets, response dedup and a usage dashboard so spend never runs away.

xiii.Common questions

Everything before you book the call.

Still missing something? Drop a note in the form below — we usually reply within a few hours.

A production-grade v1 typically takes 10–14 weeks. Simpler internal tools can land in ~3 weeks; complex multi-tenant platforms run closer to 16. We'll give you a realistic estimate within 48 hours.
We work on fixed-scope sprints or monthly retainers. Most v1 builds land between $80K–$240K depending on surface area and AI complexity. Retainers start at $14K/mo.
Yes — we'll send one over before the first discovery call. Every team member is bound by a master mutual NDA. IP, code and roadmap stay yours.
Absolutely. We embed inside your team, pair with your engineers, and hand off cleanly — augmenting a small in-house team is one of the ways we like to work.
You do. Everything. Source code, infrastructure, accounts, secrets, runbooks — handed over fully.
You can keep us on as an embedded retainer for ongoing growth engineering, platform care and AI evolution — or take the whole thing in-house. Both are great.
Yes — scale is a first-class concern, not an afterthought. Multi-tenancy, horizontal scaling, edge caching and realtime infrastructure are defaults in our stack, so we design for an order of magnitude beyond your current load.
2 build slots open for Q3 2026
Start something

Let's build the next thing you're known for.

Tell us about your product, timeline and constraints — we'll come back within a day with a concrete plan, not a brochure.

  1. 1
    Tell us the shape of itProduct, timeline, constraints — a few lines is plenty.
  2. 2
    We reply within 48 hoursA concrete plan and rough scope — from an engineer, not a sales rep.
  3. 3
    We start buildingSenior engineers in your stack, shipping a v1 in weeks.
0h
First-reply time
0%
You own the code
Secure
By design
Tell us about your project
We'll respond within 24 hours with next steps.