CognetixAI
Let's talk

AWS Consulting · AWS Training · Applied AI

Cloud architecture that ships. AI that earns its keep.

I'm Tomas Miselnicky. Through CognetixAI I help teams design, build and run on AWS, train their people to own it, and put generative AI to work where it actually helps. Hands-on, independent, in English, Spanish or Czech.

Talk through your project Explore services
  • Based in Los Angeles, USA
  • Clients in the US, Europe & Asia
  • English, Spanish & Czech

Services I work with

Services I work with: Amazon S3, CloudFront, Route 53, ACM, Lambda, Step Functions, DynamoDB, Amazon Bedrock, CloudFormation, IAM, Organizations, API Gateway and CloudWatch.

01 / Services

Three practices, one standard: it has to work in production.

Solid cloud foundations make AI possible; training makes both stick. I usually start with the AWS foundation and finish when your team can run it alone.

Core practice

AWS Consulting

Architecture, builds and clean-ups for teams that want AWS to be fast, secure and boring in the best way, with costs you can explain.

Talk about AWS consulting
  • Architecture reviews against the Well-Architected Framework
  • Migrations from on-prem, VPS or other clouds, with a rollback plan
  • Serverless builds: Lambda, API Gateway, Step Functions, DynamoDB
  • Cost optimization, right-sizing, budgets and alerts
  • Security, IAM and multi-account landing zones
  • Static sites & CDN hosting on S3 + CloudFront
  • Infrastructure as Code with CloudFormation
  • Runbooks and handover documentation
Enablement

AWS Training

Practical training built around real consoles and real mistakes, so your team leaves able to do the work, not just pass a quiz.

  • Hands-on team workshops
  • Certification prep: Cloud Practitioner, Solutions Architect Associate, Developer
  • Custom bootcamps built on your stack
  • Labs in sandbox AWS accounts, with cleanup scripts included
  • In English, Spanish or Czech, remote or on-site
Plan a training
Applied AI

AI Consulting

From "we should do something with AI" to a working system: scoped to a business outcome, built in your AWS account, with the evaluation and guardrails to trust it.

  • Generative AI strategy and use-case selection
  • Amazon Bedrock implementation: model access, inference profiles, IAM, cost
  • RAG and internal knowledge assistants
  • AI agents & automation: WhatsApp, voice, back-office
  • LLM evaluation and guardrails
Explore an AI use case

02 / Process

How I work: small steps, written down, handed over.

Four phases, each with something you keep. You can stop after any of them and still walk away with something useful.

  1. Discover

    We map goals, constraints, current accounts and spend. What matters, what doesn't, and where the risk is.

    → Written read-out & priorities

  2. Design

    A target architecture, a phased plan, a cost estimate and the trade-offs in plain language.

    → Architecture & phased plan

  3. Build

    Infrastructure as code, in your accounts, delivered in reviewable increments. No black boxes.

    → Working system in your account

  4. Enable

    Runbooks, documentation and hands-on sessions until your team owns it. The goal is that you don't need me.

    → A team that owns it

Code over click-ops. Cost is a feature. Handover by default.

  • 01If it isn't in a template, it doesn't exist.
  • 02Every design comes with a cost estimate.
  • 03Code, docs and access stay with you.

03 / Engagements

Pick the shape that fits the problem.

Scope and price are agreed in writing before anything starts. Here is how work is structured.

Good place to start

Advisory sprintFixed scope

Review & roadmap

A focused look at your architecture, security posture, costs or AI idea, ending in a prioritized written plan.

Fits: you need clarity before committing budget.

ProjectDefined outcome

Design & build

A migration, a serverless backend or a Bedrock assistant, delivered with code, docs and handover.

Fits: you know what you need built.

RetainerMonthly

Fractional cloud lead

A set number of hours each month for architecture decisions, reviews, cost check-ins and a second opinion.

Fits: you don't need a full-time cloud architect yet.

Training cohortPer group

Workshop or bootcamp

From a one-day hands-on workshop to a multi-week certification track, in English, Spanish or Czech.

Fits: you want your people to own the platform.

04 / Field notes

I run what I recommend. Here's what's in production.

Not client logos: my own systems, built and operated on AWS every day. The same patterns go into client work.

CASE_01 · mxsityIn production

A web agency fleet on S3 + CloudFront

Context

I run mxsity, a web agency that builds and hosts websites for small businesses in Mexico. More than 200 client sites need to be fast, secure and cheap to operate.

What I built

The whole fleet is served from Amazon S3 behind CloudFront, with Route 53 DNS and ACM certificates, deployed through scripted AWS CLI workflows. No servers to patch.

  • S3
  • CloudFront
  • Route 53
  • ACM
  • AWS CLI
CASE_02 · agentsIn active development

An AI-agent platform on Amazon Bedrock

Context

Small businesses lose customers to unanswered messages, missed calls and unreplied reviews. They need help that works around the clock without an IT team.

What I'm building

A platform of agents (WhatsApp agent, phone receptionist, review responder, content generator) on Amazon Bedrock with Lambda, Step Functions and DynamoDB, all defined in CloudFormation.

  • Bedrock
  • Lambda
  • Step Functions
  • DynamoDB
  • CloudFormation
CASE_03 · opsIn production

Multi-account operations

Context

Real businesses rarely live in one tidy account. Domains, distributions and AI workloads end up spread across several.

What I run

I manage multiple AWS accounts day to day, including cross-account Route 53 and CloudFront setups and Bedrock inference-profile access, so I know where the sharp edges are.

  • IAM
  • Route 53
  • CloudFront
  • Bedrock

05 / About

Builder first. Consultant second.

I'm Tomas. I build and run things on AWS for a living, and I started CognetixAI to help other teams do the same with less guesswork.

Day to day I operate a fleet of more than 200 websites on S3 and CloudFront for my agency, mxsity, and I'm building an AI-agent platform on Amazon Bedrock. That's the lens I bring to consulting and training: what holds up in production, what it costs, and what your team can realistically maintain.

I work with clients across the US, Europe and Asia, in English, Spanish or Czech, mostly remotely. From Los Angeles, early-morning calls catch the European afternoon and late-day calls catch the Asian morning.

  • IndependentNo AWS sales targets or partner incentives behind my recommendations.
  • Hands-onI write the code and the templates, not only the slides.
  • Plain languageTrade-offs explained so non-engineers can decide too.
  • HandoverEverything lives in your accounts, documented.

06 / FAQ

Good questions to ask first.

Something else on your mind? Email me and I'll answer directly.

Are you an AWS Partner?

No. CognetixAI is independent and not affiliated with Amazon Web Services. I'm not tied to AWS sales targets or partner incentives, so recommendations are based only on what fits your situation.

Will you need access to our AWS account?

For hands-on work, yes, through a least-privilege IAM role you control and can revoke at any time. I never ask for root credentials or long-lived access keys. Everything is built in your accounts; nothing is hosted on mine.

Where does our data go in an AI project?

It stays under your control. I build on Amazon Bedrock in your AWS account, under your IAM policies; AWS doesn't use your prompts or outputs to train its base models. Region choice (including cross-region inference profiles) and data boundaries are documented as part of the design.

Can training be delivered in Spanish or Czech?

Yes. Workshops, bootcamps and certification prep are available in English, Spanish or Czech, remote or on-site. I explain AWS terminology in your team's language so they can use the official docs with confidence.

We're curious about AI but don't know where to start.

That's a great place to start. An advisory sprint identifies one or two use cases with a clear purpose, checks the data you have, and estimates what a Bedrock prototype would take.

How do pricing and getting started work?

Send an email with a few lines about your situation and we'll set up a short call. If there's a fit, you get a written proposal with scope, timeline and price: fixed for sprints and projects, monthly for retainers, per group for training.

Next step

Tell me what you're building. I'll tell you how I'd run it on AWS.

A few lines are enough: what you have today, where it hurts, and what you'd like to be true in three months. I reply personally.

Email me about a project Ask about training
tom@cognetixai.com