Berlin / Potsdam · Technical Product Owner
LinkedIn ↗

Technical Product Owner.
I turn operational problems into clear product requirements and working AI, data and automation products.

I work across discovery, workflow design, prioritisation, implementation and delivery—connecting business needs with technical execution.

Product discovery Requirements & prioritisation Technical delivery Validation & iteration

Open to Product Owner, Junior Product Manager, Product Operations, Implementation and AI/Data Product roles.

Professional backgroundFormer Data Analyst
Based inBerlin / Potsdam
EducationBSc Computer Science
Current focusProduct ownership + delivery
LocationBerlin / Potsdam
EducationBSc Computer Science
ExperienceFormer Data Analyst
PortfolioFive product builds
01 · Product delivery

From problem to working product.

I identify the operational problem, define the user and business outcome, translate it into requirements, work through implementation and validate whether the result solves the original problem.

Discover the problem
Define users and outcomes
Translate needs into requirements
Prioritise the product scope
Coordinate implementation
Test, learn and iterate
02 · Professional work

Professional evidence of product delivery.

Work at the intersection of business requirements, operational users, data and implementation.

kloeckner.i GmbH · Berlin · 2024–2025

Turning a recurring document process into a reliable data workflow.

DiscoverUnderstood a recurring document and data workflow that consumed analyst time
DefineMapped the process and translated operational needs into automation and validation requirements
DeliverImplemented API ingestion, Python/SQL processing, validation and monitoring views
ValidateChecked outputs with operational stakeholders and refined the workflow
Outcome15+ hours/week saved · 28% fewer errors
TechnologyPython · SQL · BigQuery · Power BI · APIs
Ripples Pure Water · 2020–2021

Operational data for better sales, inventory and distribution decisions.

Standardised customer data, improved recurring reporting and helped turn operational questions into clearer KPIs and replenishment decisions.

Customer records500+ standardised
Recurring reporting25+ staff hours/month saved
Stockouts40% reduction contribution
03 · Selected projects

Selected product work.

Products I have founded, delivered for clients, prototyped and tested across AI, data and operational workflows.

Currently building
Currently building
01 · My active product · Private event networking

MeetBook → Nimo

Nimo is a private event contact book that helps people capture who they meet, remember the context behind each connection and follow up after the event.

Target user: People making connections at eventsProblem: Contact tools lose the context behind a meetingMy role: Founder · discovery · positioning · prioritisation · deliveryDeliverable: Launch-ready private contact and follow-up product

Originally developed as MeetBook, the product is now evolving into Nimo ahead of its October 2026 launch.

Launching 15 October 2026
Nimo home interface with a live event, recent contacts and follow-up context
View case study6 screens →
Private repository
ExpoTypeScriptAI notesGlance
Currently building · Freelance client project
02 · Client application · Local services

Bidly

A local-services marketplace where customers post a job, nearby providers bid, and the customer chooses using price, ratings, reviews and availability.

Target users: Namibian customers and service providersProblem: Local-service requests and offers are fragmentedMy role: Freelance application development and technical deliveryDeliverable: Working customer and provider application flows
Namibia
Bidly home interface for posting a local service request
View case study6 screens →
Private client projectPrivate repository
ExpoTypeScriptSupabaseSelecomPay
Selected projects and experiments
03 · AI contract-intelligence product · Working prototype

ContractGuard AI

A working prototype for teams that need to turn uploaded contracts into structured deadlines, risk signals and reviewable information.

Problem: Manual review makes key contract dates easy to missMy role: Product design · workflow definition · AI implementationDeliverable: Reviewable extraction and contract-tracking workflow
ContractGuard AI command center showing an active contract, upcoming deadline and risk alert
View case study7 screens →
View live product ↗Private repository
ReactTypeScriptSupabasePostgreSQL
04 · Enterprise AI-agent assurance product · Functional prototype

ACPT

An assurance product for enterprise teams that need workflow monitoring, investigation and audit evidence around AI-agent outcomes.

Problem: Technical execution success does not prove the right business outcomeMy role: Product strategy · assurance architecture · workflow design · implementationDeliverable: Phase 1 assurance and investigation prototype
ACPT assurance overview monitoring an invoice agent, workflow activity and incident health
View case study7 screens →
View live product ↗Private repository
n8nAI assuranceAudit evidence
05 · Reproducible data-research pipeline · Public technical evidence

ORB Market Research Pipeline

A research pipeline for analysts exploring five-minute QQQ market data around the New York opening range.

Problem: Market hypotheses require consistent, validated research inputsMy role: Research design · ingestion · validation · exploratory analysisDeliverable: Reproducible research foundation, not a profitability claim
Notebook heatmap exploring no-1R outcome probability across opening-range and volume regimes
View case studyResearch evidence →
View repository ↗
PythonJupyterAlpaca APIPandas
  • Python
  • SQL
  • Power BI
  • TypeScript
  • React
  • React Native
  • Expo
  • Supabase
  • PostgreSQL
  • REST APIs
  • n8n
  • OpenAI
  • Claude
  • Codex
  • Cursor
  • GitHub
  • Jupyter
  • Data Validation
  • Workflow Automation
04 · AI-assisted implementation

How I work with AI.

I use AI coding tools for speed, but the system around them matters more: context before execution, bounded tasks, acceptance criteria, testing, guardrails and human judgement over architecture and product decisions.

Context Plan Build Test Review Human decision
Python
SQL
TypeScript
React
OpenAI
Claude
n8n
REST APIs
Data validation
Testing
Human judgement
AI assistance · connected by process · controlled by judgement
05 · Product ownership capabilities

From discovery through delivery.

Capabilities demonstrated through professional workflow improvement, founder-led product work, client delivery and technical prototypes.

Product discovery

Understand the problem

User and stakeholder needs, problem definition, workflow mapping and requirement clarification.

Product definition

Shape the right scope

Product requirements, user stories, acceptance criteria, scope decisions and prioritisation.

Delivery

Move work to completion

Roadmap planning, backlog organisation, cross-functional coordination, testing and iteration.

Data and AI

Design informed workflows

Data analysis, AI workflow design, automation and outcome validation.

Technical collaboration

Connect product and engineering

APIs, integrations, data models, frontend/backend understanding and technical documentation.

Launch and learning

Validate and improve

Product validation, feedback loops, metrics and iteration planning.

06 · Thinking in public

Posts, lessons & experiments.

Public thinking from LinkedIn — read it here, then open the thread if you want the rest.

LinkedIn profile ↗
Highlight · SummerUP

Validate before you finish building.

I went into the hackathon thinking the hard part was building fast. I left realizing that when software can be built in days, you can also build the wrong thing in days.

Open on LinkedIn ↗
TP
Tanatswa Phil Muganga AI Automation Engineer · Berlin
in

I went into SummerUP thinking the hard part was building fast. I left realizing that building fast might actually be the dangerous part. Because when software can be built in days, you can also build the wrong thing in days.

One lesson stuck: validate before you finish building. Talk to potential customers before the product is done. Ask questions. Get on calls. Understand how they deal with the problem today and whether what you’re building actually matters to them.

The idea I was working on was ACPT — governance and control for AI agents. Being surrounded by people from different industries changed the question from “Can I build this?” to “Should I build this?”

LinkedIn · 3d Read post ↗
TP
Tanatswa Phil Muganga AI Automation Engineer · Berlin
in

Two days ago I attended the Engineering AI Together Unconference at the Merantix AI Campus in Berlin — speaking with AI engineers, sharing VoiceOps, and getting honest feedback.

The conversations moved from benchmarking, multi-agent workflows and security into product decisions. One person asked why VoiceOps should be desktop-first when most people already use voice assistants on their phones. A fair question — and it made me think more carefully about the environment the product is actually meant to be used in.

LinkedIn · 3w Read post ↗
TP
Tanatswa Phil Muganga AI Automation Engineer · Berlin
in

Last Friday I joined Build Fridays in Berlin, hosted by AI BEAVERS, to advance the product direction and technical foundation for VoiceOps.

I used the session for focused product discovery with engineers — workflows, pain points, workarounds, privacy concerns, desktop vs mobile. The result was not a feature list. It was a clearer product direction, a validated MVP focus, and an architecture aligned with the highest-value user needs.

LinkedIn · 2w Read post ↗
07 · Contact

Let’s work on
something difficult.

If you’re hiring for applied AI, digital systems or automation, I’d be happy to talk.

tanamuganga16@gmail.com +49 152 92606260 LinkedIn ↗
TanaPhil · Tanatswa Phil Muganga · Berlin / Potsdam · © 2026.