Software development · DevOps · AI systems

I build reliable systems and practical AI products.

I’m Kutlo Sekwamote, a software developer working across production engineering, cloud infrastructure, automation, machine learning, and AI-powered products. My work spans national-scale platforms, business systems, and independent SaaS products built to solve specific operational problems.

Explore featured work ↓ Work with me ↗
Gaborone, Botswana 5+ years in software engineering Production systems & infrastructure
engineering_profile.json
Current profile

Software Developer
with a systems mindset.

Full-stack foundations, DevOps execution, and a growing focus on deploying useful AI into real products.

03 Live independent AI products
National Scale of current platform work
CI/CD Automation and release engineering
24/7 Reliability and observability thinking
Build
→
Deploy
→
Observe
Selected organisations & clients
BITRI Thito Holdings Risk Engineers Botswana BOPRITA Botho
About

Engineering across the full lifecycle.

My experience crosses software development, cloud infrastructure, automation, and operational reliability. I currently contribute to national-scale systems at the Botswana Institute for Technology Research and Innovation, where dependable environments, repeatable deployments, monitoring, and long-term maintainability matter as much as the application code itself.

I design CI/CD pipelines, maintain containerised environments, configure reverse proxies and Linux servers, implement observability with Prometheus and Grafana, and work with databases that support production applications. My background in full-stack engineering means I can follow a system from interface and API design through deployment, infrastructure, monitoring, and support.

Alongside that work, I am building applied AI products and strengthening my machine-learning practice. I am most interested in AI that has a clear product boundary: structured assessments, retrieval-grounded answers, document intelligence, transcription, workflow automation, and systems that disclose their tradeoffs.

I care less about adding “AI” to a product and more about making the product measurably more useful, reliable, and honest.
Featured products

Three products. Three distinct engineering problems.

Each product was designed around a specific constraint: serverless execution, grounded generation, or consistently structured LLM output.

verbaitim-ai.vercel.app
Verbaitim AI transcription product homepage
01 / AI transcription

Verbaitim

Accurate, exportable transcription engineered around a hard constraint: building a low-cost product entirely on free-tier infrastructure without routing large media files through serverless API limits.

  • Direct browser-to-Supabase uploads through signed URLs; Next.js only orchestrates job state.
  • Groq-hosted Whisper Large-v3 for fast transcription, with Firebase token verification on API routes.
  • Timestamp-based TXT, SRT, VTT, and DOCX exports, plus Gemini translation on demand.
  • Speaker separation is a documented gap heuristic—a disclosed tradeoff, not a hidden guess.
Next.js Supabase Storage Groq Whisper Firebase Auth Gemini
Visit Verbaitim ↗
citebox.vercel.app
Cited retrieval-grounded chat product homepage
02 / Retrieval-grounded chat

Cited

A document-grounded support widget that lets a business upload its own material and embed a branded chat experience on another website through a single script tag.

  • Answers are generated from indexed source passages and returned with inline citations.
  • The customer integration surface is one script tag rather than a custom frontend project.
  • Branding and accent colour are generated per account at embed-code creation time.
  • Built for policies, FAQs, manuals, and other documents where traceability matters.
RAG Embeddings Document chunking Embeddable widget Inline citations
Visit Cited ↗
gauge-ai.vercel.app
Gauge AI readiness assessment homepage
03 / Structured LLM output

Gauge

An AI-readiness assessment that scores a business across four independent dimensions instead of hiding the result behind one decorative number.

  • A six-question intake covering the business, team, bottlenecks, and desired future state.
  • Separate scores for data, team, process, and tooling make the result actionable.
  • The model must produce consistently structured output for the breakdown, quick wins, and roadmap.
  • The result includes three immediate actions and a concrete six-month implementation plan.
Structured generation Schema validation Assessment logic Roadmap generation SVG visualisation
Visit Gauge ↗
Expertise

What I work across.

A broad engineering foundation, with deeper focus on infrastructure, automation, production reliability, and useful AI systems.

</>

Software development

Full-stack products, APIs, dashboards, responsive web applications, and maintainable application architecture.

ReactNext.jsNode.jsSpring BootTypeScript
CI

DevOps & automation

CI/CD pipelines, automated releases, containerised environments, reverse proxies, backups, and infrastructure scripting.

JenkinsDockerSwarmNginxLinux
☁

Cloud infrastructure

Production deployments designed around reliability, security, scalability, networking, and practical cost control.

AWSAzureFirebaseVPCRoute 53
AI

AI product engineering

Retrieval systems, document intelligence, assistants, transcription, workflow automation, and model integrations.

RAGEmbeddingsLLM APIsFastAPILangChain
ML

Machine learning

Applied ML foundations across supervised learning, deep learning, vision, sequence models, transformers, and evaluation.

PythonTensorFlowPyTorchScikit-learnPandas
DB

Databases & observability

Database design and operations, migrations, backups, performance troubleshooting, monitoring, alerting, and reporting.

MariaDBPostgreSQLMSSQLPrometheusGrafana
Selected client systems

Built for real operational work.

A selection of business systems designed to replace fragmented processes with clearer, more trackable digital workflows.

01

Loan application platform

Thito Holdings

A customer-facing application for digital loan submission, document capture, status tracking, approvals, and a smoother path from application to decision.

02

Property management application

Sycamon

A central system for property records, tenants, listings, requests, documentation, and the day-to-day workflows behind a property portfolio.

03

Internal loan tracking

Thito Holdings

A structured operations dashboard for loan records, stages, approvals, staff follow-ups, and accountability.

04

Helpdesk & ticket management

Thito Holdings

A Kanban-based support workflow that makes ownership, priority, progress, and turnaround visible from one place.

05

Risk management services

Risk Engineers Botswana

A client-facing platform for risk strategy, engineering consultancy, training, reporting, modelling, and mitigation services.

Experience

From support to full-stack to production systems.

The progression has been practical: understand the user, build the application, automate delivery, and take responsibility for how the system behaves in production.

APR 2023 — PRESENT

Software Developer — DevOps

Botswana Institute for Technology Research and Innovation

Contributing to national-scale platforms through application engineering, environment design, containerised deployments, CI/CD, monitoring, infrastructure operations, and production reliability.

JAN 2020 — APR 2023

Full-stack Engineer

Thito Holdings

Built and maintained customer-facing and internal business systems across loan processing, operational dashboards, support workflows, APIs, and application deployment.

FEB 2019 — JAN 2020

Technical Support

IT operations

Installed and configured hardware, software, networks, printers, and user systems while supporting day-to-day technical operations and troubleshooting.

JAN 2018 — DEC 2018

Lecture Assistant

University of KwaZulu-Natal

Supported teaching and technical learning, helping students work through foundational computing concepts and practical tasks.

Tools I use to move a system from idea to production.

Technology choices change with the problem. The constant is designing for maintainability, observability, and a realistic operating environment.

JavaScript TypeScript React Next.js React Native Node.js Express Spring Boot Python FastAPI Docker Docker Swarm Jenkins Nginx Linux AWS Azure Firebase Supabase MySQL MariaDB PostgreSQL MSSQL MongoDB Prometheus Grafana Alertmanager TensorFlow PyTorch Scikit-learn LLM APIs Vector databases
Writing

Notes from building and deploying.

Practical articles on software design, cloud deployment, and application architecture.

Software design

Simplify Your Code: The Power of the KISS Principle in JavaScript

Why simple, understandable code is easier to maintain, debug, and extend than unnecessary complexity.

Read on Medium ↗ Cloud infrastructure

How to Point Your Domain to an EC2 Instance Using Route 53

A practical deployment walkthrough connecting domain configuration, DNS, and a production EC2 environment.

Read on Medium ↗ Frontend architecture

Protected Routes in React Using React Router and a HOC

A pattern for controlling access to pages and components through reusable authentication logic.

Read on Medium ↗
Contact

Have a system to build, improve, or stabilise?

Reach out about software development, infrastructure, deployment automation, production reliability, AI product engineering, or technical collaboration.

Email ksekwamote@gmail.com
LinkedIn linkedin.com/in/kutlosek
Technical channel YouTube / Koding 101
Location Gaborone, Botswana