Organizational Psychology · AI Systems · People Ops

I design systems
that make
organizations
work better.

Behavioral research practitioner and systems designer. I close the gap between how people perform and how organizations support that performance — through measurement tools, learning systems, and AI-augmented workflows.

About

I'm Ovi — an organizational psychology practitioner based in Malawi (Remote · UTC+2), specializing in the intersection of behavioral science, systems design, and AI workflow engineering.

At DeepThought CultureTech, I built psychological safety interventions, performance measurement tools, and AI-assisted operational systems across 40+ people and 7 distributed labs — reducing execution friction while making the work environment genuinely better for the people inside it.

Available for remote roles and fractional engagements with async-first, mental-health-forward organizations.

View DeepThought Leadership Portfolio →

Psych SafetyPeople Ops AI WorkflowsResearch Learning Design
40+
People impacted
7
Labs covered
40%
Efficiency gain
Content speed-up

Selected Work

05 Projects
Case Study · 01
Psychological Safety Intervention System
A full diagnostic-to-intervention behavioral program addressing inferiority complex and imposter syndrome across a distributed cohort of 40+ people.
Behavioral DesignPeople ProgramsOrg Psychology
Case Study · 02
Psychology-Grounded Performance Intelligence Tool
A six-dimension achievement measurement system — each anchored in published theory — designed to surface internal signals that predict disengagement before any external metric drops.
People AnalyticsMeasurement DesignOrg Psychology
Case Study · 03
Clarity Arena: Socratic Peer Learning System
A structured dialogue methodology using Socratic-Feynman questioning to build real-time reasoning and knowledge transfer across four business domains simultaneously.
Learning DesignFacilitationPsych Safety
Case Study · 04
Unicorns Assemble: Team Engagement Architecture
A behavioral engagement system combining servant leadership rituals, healing hackathons, and AI-powered feedback loops to shift distributed teams from task execution to leadership identity.
Engagement DesignBehavioral DesignAI Systems
Case Study · 05
Sprint Review System: AI-Augmented Reporting Infrastructure
A complete reporting infrastructure rebuilt using AI-assisted prompts, PBAN communication design, and custom scoring — cutting cycle time by 40% across 7 distributed labs.
AI WorkflowsPrompt EngineeringOps Design
Case Study 01 · Organizational Psychology · DeepThought CultureTech

Designing a Psychological Safety
Intervention from Scratch

When low self-esteem and imposter syndrome were quietly eroding team performance, I didn't send a motivational message. I built a diagnostic-to-intervention system grounded in behavioral psychology — and ran it across a distributed cohort of 40+ people.

Org: DeepThought CultureTech Ventures
Timeline: Feb – Oct 2025
Setting: Fully Remote · 7 Labs
Behavioral DesignPeople ProgramsPsych SafetyLearning SystemsOrg Psychology
40+
Participants
6
Form Questions
4
Workflow Stages
≥70%
Target Satisfaction
9
Frameworks Applied
The Problem

Performance was dropping. The cause wasn't capability — it was psychology.

Across multiple labs, a pattern was emerging: capable people consistently underperformed, avoided stepping forward, and disengaged from growth opportunities. The surface-level read was "low motivation." The root cause was structural — inferiority complex, imposter syndrome, and social comparison anxiety were operating unchecked inside the organization's culture.

Using the Cynefin framework, I classified this as a complex problem — not a compliance issue or a knowledge gap. There was no clear cause-and-effect chain. It required iterative probing, not a top-down solution. It needed a designed system.

Reframe

When engagement dropped, I didn't assume the issue was "lack of interest." I mapped the ecosystem: Were task assignments aligned with learner readiness? Was measurement reinforcing the right behaviors? Were reflection loops strong enough to create belief shifts?


My Role

I identified the leverage point and owned the full design cycle.

I was operating as Micro Lead and Team Lead — not a senior manager with authority to mandate change. I had influence without positional power, which meant every intervention had to be designed to be genuinely wanted, not enforced. I proposed, designed, tested, and iterated the entire initiative.


The Approach

Four stages. Diagnostic first, activity second.

01

Anonymous Diagnostic Form

Designed a 6-question psychological safety form using metaphor-driven language instead of clinical jargon. "Do you ever feel like everyone else is succeeding and you're falling behind?" — not "Do you experience inferiority complex symptoms?"

Each question mapped to a specific psychological construct with documented reasoning. The Q3 swimming pool metaphor alone maps to 7 distinct constructs — imposter syndrome, social anxiety, decision paralysis, avoidance — without a single clinical term.

02

Thematic Clustering & Activity Design

Responses were clustered into themes, not read literally. From these I designed a bank of growth activities — gamified, time-boxed (10–30 minutes), with clear boundaries between private and shareable. Guiding principle: structured sharing of insights, never wounds.

03

Episodic Rollout with AIDCA Nudges

Deployed as a series (Season 1: 3 Confidence Builders) rather than a single event. AIDCA-structured nudges replaced reminders: "Do you feel like everyone else is ahead? What if you weren't alone in this?" Pull, not push.

04

Feedback Loops & Iteration Framework

Post-activity reflections with before/after ratings. Decision framework documented in advance — if resistance increases → simplify; if activities underperform → rotate from the bank. KPIs set before launch: ≥30 form responses, ≥20 participants, ≥70% satisfaction, ≥50% retention at 1 month.

Design Principle

"Cringe" meant forced vulnerability, juvenile games disconnected from growth, surface-level cheerleading, and ambiguous expectations. Every activity was tested against this definition before deployment.


Frameworks Applied

Not intuition. Applied theory with documented reasoning.

Cynefin FrameworkSelf-Determination TheorySelf-Efficacy (Bandura)Golden Circle (Sinek)AIDCA ModelSystems ThinkingDesign ThinkingSocratic QuestioningFeynman Technique

Outcomes

What changed — and what the data showed.

Psychological safety infrastructure installed

Where previously there was no structured mechanism for addressing self-esteem and identity challenges, the organization now had a documented, repeatable system — form, activity bank, rollout cadence, iteration framework, and KPIs.

Observable shifts in participation and language

Self-reflection language in sprint reports shifted from output-oriented to growth-oriented. Stakeholders reported improved emotional safety and peer collaboration across learning cohorts.

Replicable model, not a one-off event

The activity bank, form structure, and iteration logic persist beyond any individual facilitator. The design investment compounds over time.


Reflection

"I shifted from effort-driven execution to leverage-driven systems thinking. Instead of completing tasks manually, I began designing prompts, workflows, and environments where clarity, confidence, and morale made performance sustainable. The shift was from enforcement to institutional clarity and emotional infrastructure."

Case Study 02 · People Analytics · DeepThought CultureTech

Building a Psychology-Grounded
Performance Intelligence Tool

Sprint reports measure what people did. UBS scores measure how people behaved. Neither measured how people felt about their own growth — the internal signal that predicts disengagement before any external metric shows a problem.

Org: DeepThought CultureTech Ventures
Timeline: Feb – Oct 2025
Setting: Fully Remote · 40+ People
People AnalyticsPerformance SystemsMeasurement DesignPeople OperationsResearch
10
Likert Items
6
Psych Constructs
5
Process Steps
40+
People Covered
2wk
Sprint Cadence
The Problem

The organization was measuring outputs. It wasn't measuring what creates them.

UBS scores tracked behavioral quality, sprint reports tracked execution, supervisor feedback tracked performance. None of it measured whether people actually felt they were growing. By the time a UBS score drops, disengagement has already been happening for weeks. I designed a tool to catch that signal earlier.

Core Insight

External performance data and internal psychological experience measure different things. You need both. The Achievement Scorecard bridges that gap — not replacing existing metrics, but completing the picture.


What I Built

Six dimensions. Each grounded in published psychological theory.

Pride & Self-Recognition

Anchored in Self-Determination Theory (Deci & Ryan). Pairs felt pride with conscious internalization of growth — preventing external validation dependency and building identity development.

Confidence & Self-Belief

Anchored in Bandura's Self-Efficacy Theory. Measures both current confidence and recognition of past-to-present progress — anchoring self-belief in evidence, not feeling alone.

Growth & Learning

Anchored in Dweck's Growth Mindset and Kolb's Experiential Learning Cycle. Quantitative skill check paired with open-ended identity reflection captures both tangible and aspirational growth.

Contribution & Impact

Anchored in Hackman & Oldham's Job Characteristics Model. Assesses durability of contribution and generativity — not just "did I contribute?" but "did my work open possibilities?"

Energy & Alignment

Anchored in Work Engagement Research (Kahn; Maslach & Leiter). Energy and values alignment are the early warning signals for burnout — this dimension catches misalignment before disengagement.

Recognition & Validation

Anchored in Social Identity Theory. Measures whether individual effort is being mirrored back by peers and mentors — the social reinforcement that converts individual pride into belonging.


Process Design

Five-step flow from data collection to intervention.

01

Data Collection — Every 2-Week Sprint

10 Likert-scale items + 2 open-ended reflections at sprint end. Open-ended questions use expressive writing methodology (Pennebaker) to capture nuance scales cannot — the texture of someone's growth, the specific moment something clicked.

02

Aggregation & Cross-Analysis

Scores averaged per sprint, rolled into monthly trends. Cross-referenced with UBS scores, sprint reports, and peer feedback to identify correlations — not isolated data points.

03

Three-Level Analysis

One instrument, three layers of insight with no duplication of effort:

  • Individual → personal nudges and targeted support
  • Team → systemic patterns and ritual design
  • Org-wide → KPIs for leadership reporting
04

Tiered Intervention Logic

Response protocols defined in advance: individual drop → mentorship or check-in; team gap → refine rituals, adjust scaffolding; org-wide signal → escalate with data, not anecdote.

05

Dual Feedback Loop

Program Managers receive dashboards across all six dimensions. Participants receive personalized reflection nudges. Both sides informed — this is the difference between surveillance infrastructure and developmental infrastructure.

Governance Design

The SOP includes full governance — who owns completion (Program Managers), how data auto-aggregates (ART Lab/PDGMS system), frequency (every 2-week sprint, monthly review), and integration with existing governance meetings. A measurement tool without governance is just a form.


Outcomes

What this changed at the organizational level.

Closed the gap in measurement infrastructure

The organization previously had no instrument measuring the internal psychological experience of achievement. This tool filled that gap with a theoretically grounded, operationally integrated system.

Early disengagement signals now catchable

By measuring energy, alignment, and confidence at sprint cadence, the system surfaces early warning signals before they manifest as performance drops — giving time to intervene effectively.

Scalable across cohorts without additional design work

The SOP, rubrics, governance structure, and intervention logic are documented and repeatable. Any new cohort onboards without rebuilding — the design investment compounds.


What This Demonstrates

"I don't just design tools — I design systems with theoretical grounding, operational integration, governance structures, and feedback loops built in from the start. The scorecard is not a survey. It is a piece of organizational infrastructure."

Case Study 03 · Learning Systems · DeepThought CultureTech

Clarity Arena: A Socratic Dialogue
System for Distributed Teams

Most team meetings produce the illusion of understanding. Clarity Arena was designed to produce the real thing — structured Socratic-Feynman dialogue sessions where participants had to explain, defend, and stress-test their thinking in real time, across four business domains simultaneously.

Org: DeepThought CultureTech Ventures
Timeline: Sep – Oct 2025
Setting: Fully Remote · 40+ Participants
Learning DesignFacilitationSocratic MethodKnowledge SystemsCommunication Design
Sessions/week
4
Business Domains
80%
Participation Target
40+
Participants
The Problem

People were executing without truly understanding. That gap was expensive.

Teams worked across four domains — BIZ, PMF, GTM, and OPS. Each required different reasoning. But the organization had no mechanism to assess whether participants understood the logic behind their work, or were simply completing tasks.

The symptoms: shallow sprint reflections, inability to explain decisions under pressure, low-quality peer feedback, and the same problems recurring because root causes were never interrogated. Execution without comprehension is just expensive motion.

Design Challenge

The system needed to build genuine understanding — not performance of understanding. This meant creating conditions where participants had to explain concepts in their own words, respond to probing questions, and defend their reasoning. The Feynman Technique wasn't a reference — it was the operating principle.


What Clarity Arena Is

A structured peer-reasoning forum. Not a lecture. Not a Q&A. A live stress test of understanding.

Clarity Arena ran three times per week across all four business domains. Topics were drawn from real organizational challenges — not hypothetical exercises. Participants presented positions, responded to Socratic questioning, and demonstrated reasoning depth. Examples from sessions I designed and facilitated:

BIZ Domain

"How is DT Avahana going to help organizations in their execution? What is the importance of feedback and how do you provide it to help teams improve?"

OPS Domain

"Why is it important to escalate issues, and how does that impact the organization? How can we create a winning culture and ensure engagement with L&D activities?"

PMF Domain

"How can we achieve resonance and increase CTR by ideating better and experimenting with content? What parameters matter for a QA Agent in content review?"

GTM Domain

"How can we maintain quality during a transition phase? What frameworks should we follow to ensure deliverable quality remains high?"


My Role

Topic curator, session facilitator, and rubric designer.

I designed the topic bank feeding each week's sessions — translating live organizational challenges into dialogue-ready questions with enough complexity to require genuine reasoning, not shallow recall. I also facilitated sessions directly, applying Socratic questioning: probing surface-level answers, redirecting participants toward deeper causal reasoning. The facilitation model was never tell, always ask.

Facilitation Principle

The goal was not to expose what people didn't know — it was to help them discover what they actually thought. Questions were designed to surface assumptions, not humiliate. Psychological safety in the room was a prerequisite for intellectual honesty in the dialogue.


The Methodology

Socratic method meets Feynman Technique, operationalized for a distributed team.

01

Topics extracted from live organizational problems

Topics were not invented — they were extracted from real blockers, strategy gaps, and decisions the organization was actively facing. Participants reasoned about their actual work, not simulations. It raised the stakes and the relevance simultaneously.

02

Pitch-and-probe structure

Participants pitched a position on the assigned topic. The facilitator then probed using Socratic questioning: "Why?" "What would have to be true for that to work?" "What's the strongest counterargument?" "Explain that to me as if I had no context." No position was accepted at face value.

03

Peer reflection and cross-domain synthesis

After each session, participants submitted reflections linking the topic to their own lab's work — creating cross-domain knowledge transfer and building shared mental models across teams that had previously operated in silos.

04

UBS integration and accountability

Participation, pitch quality, and reflection depth were tracked through the Unicorn Behaviour Score system. Clarity Arena wasn't an optional enrichment activity — it was integrated into the performance measurement infrastructure.


Outcomes

What the system changed.

Measurable improvement in reasoning depth

Sprint reflections submitted after Clarity Arena sessions showed higher causal reasoning quality — participants moved from describing what happened to analyzing why it happened and what should change.

Cross-domain organizational intelligence built

By running sessions across BIZ, PMF, GTM, and OPS simultaneously with shared reflection, the organization built shared mental models across teams that had previously operated in silos.

Psychological safety as precondition, not byproduct

Designing for safety-first dialogue created a feedback loop: safer sessions produced more honest reasoning, which produced better organizational decisions downstream.


Reflection

"The hardest part of facilitation is not asking good questions — it's resisting the urge to answer them. Every time I held the silence after a probing question, I was creating space for someone to discover something they couldn't have been told."

Case Study 04 · Engagement Design · DeepThought CultureTech

Unicorns Assemble:
Building Team Identity
in a Distributed Org

Getting distributed people to show up to optional growth activities is a behavioral design problem, not a scheduling one. Unicorns Assemble was the engagement architecture built to solve it — combining servant leadership rituals, healing hackathons, AI-powered feedback loops, and a behavioral scoring system that made participation visible and meaningful.

Org: DeepThought CultureTech Ventures
Timeline: Sep – Oct 2025
Setting: Fully Remote · ART Lab & LIS Lab
Engagement DesignBehavioral SystemsServant LeadershipAI WorkflowsTeam Culture
90%
Participation Target
4+
Hackathons/Quarter
2
AI Tools Deployed
40+
People Covered
The Problem

People knew L&D mattered. They weren't showing up anyway.

DeepThought's learning ecosystem — Socratic dialogue sessions, Leadership Development Intensives (LDIs), PDGMS sessions — was well-designed on paper. The bottleneck was adoption. Participation rates hovered below targets, not because the activities lacked value, but because the behavioral architecture around them was weak.

This is a classic organizational psychology problem: knowing something is good for you doesn't reliably produce the behavior of doing it. The gap between intention and action required a designed intervention, not a reminder. The question: how do you make growth feel like belonging, not obligation?

Design Reframe

The problem wasn't motivation — it was that participation in growth activities had no social visibility, no recognition infrastructure, and no identity anchor. People weren't engaging because engagement felt invisible and unrewarded. The fix wasn't incentives — it was meaning-making architecture.


The System

Four interlocking components that made growth visible, social, and identity-linked.

01

Healing Hackathons — proactive emotional infrastructure

Rather than waiting for burnout to surface before intervening, healing hackathons were scheduled proactively — minimum 4 per quarter. Each targeted a specific emotional growth dimension: self-esteem, agency, resilience, or transcendence. Pre/post tracking measured real shifts, not just attendance.

The design principle: emotional immunity is built through practice, not crisis management.

02

Servant Leadership Recognition System

Servant leadership was defined operationally — not as a personality trait, but as specific visible behaviors: peer-support logs, reflections highlighting others' growth, mentoring actions. These were tracked and published. The first formally recognized Servant Leader in the organization was identified through this system.

Making leadership visible created a social mirror — others could see what it looked like in practice, not just theory.

03

AI-Powered Feedback Infrastructure

Two GPT tools were deployed: a General Feedback GPT (embedded in orientation to surface real-time developmental feedback) and an Observation GPT (answering organizational philosophy, mission, and career questions on demand). Feedback shifted from a periodic event into a continuous ambient resource — available whenever a participant needed it, without requiring facilitator availability.

04

UBS Behavioral Scoring Integration

The Unicorn Behaviour Score made participation quantifiable and transparent. Attendance, reflection quality, and pitch participation across LDIs, SD sessions, and PDGMS were weighted and scored out of 100. Not punitive — a visibility mechanism. People could see exactly where their engagement stood and what growth in each dimension looked like.


My Contribution

I operated at the intersection of all four components — and became the first person the system recognized.

Across ART Lab and LIS Lab, I contributed to hackathon content design, the servant leadership recognition framework, and facilitation of Socratic dialogue sessions. I was also formally recognized as the first Servant Leader identified by the organization — a signal that the behavioral infrastructure I helped design was working, because it surfaced me as an example before I knew I was being observed for it.

What This Required

Designing an engagement system while also being a participant in it required careful separation of roles — designing for the whole cohort, not optimizing for your own visibility. The credibility of a recognition system depends entirely on it appearing to work for everyone, not just the designers.


Outcomes

What the system produced.

Orientation redesigned as transformation, not evaluation

The 3-day orientation process was rebuilt around transformation goals — every candidate, hired or not, was designed to leave with higher agency. Completion rate target: 70%+, up from a ~50% baseline.

Servant leadership made observable and organizational

What was previously an abstract cultural aspiration became a tracked, recognized, and publicly visible behavior pattern — a living record of the culture being built.

Proactive healing culture established

The shift from reactive intervention (responding to burnout after it surfaces) to proactive emotional infrastructure (running hackathons on schedule regardless of visible crisis) is a systems-level change that prevents compounding disengagement.


Reflection

"Being recognized as the first Servant Leader wasn't something I designed for myself — it was the system working. That's the difference between performative culture and designed culture: one produces the behavior you perform, the other produces the behavior you forgot you were doing."

Case Study 05 · AI Workflow Design · DeepThought CultureTech

Rebuilding the Sprint Reporting
Infrastructure with AI

The organization's reporting system was producing friction at every stage — delayed submissions, low-quality reflections, and a review cycle that took 6–8 days and still left decision-makers without clear data. I rebuilt it using AI-assisted prompts, structured communication design, and a scoring system that made quality visible. Cycle time dropped to 3–4 days. Submission efficiency improved by 40%.

Org: DeepThought CultureTech Ventures
Timeline: Feb – Oct 2025
Setting: Fully Remote · 7 Labs · 40+ People
AI WorkflowsPrompt EngineeringOps DesignSystems ThinkingSOP Design
40%
Efficiency Gain
6→3
Days per Cycle
7
Labs Covered
3–6hrs
Saved/Team/Week
The Problem

The reporting system was producing noise, not signal.

Across 7 distributed labs, the sprint reporting process was the organization's primary visibility mechanism. In theory: the system's nervous system. In practice: consistently delayed, inconsistently formatted, and insufficient for decision-making.

Root causes were structural, not motivational. People weren't submitting late because they didn't care — the process was cognitively expensive, the format was ambiguous, the feedback loop was slow, and there was no clarity on what "good" looked like. The system was generating effort without generating insight.

Diagnosis

Using root cause analysis, I mapped the failure points: unclear submission expectations, no standardized format, no AI assistance to reduce cognitive load, no scoring rubric to define quality, and a review cycle requiring too many back-and-forth touchpoints. Each was a solvable design problem, not a people problem.


What I Built

Three interlocking systems that removed friction at every stage of the reporting cycle.

01

AI-Assisted Prompt Architecture (7 Labs)

A library of context-enriched AI prompts — one set per lab, tailored to each lab's specific deliverable types, terminology, and stakeholder expectations. Each prompt was engineered to:

  • Pre-populate structural scaffolding so writers started with a skeleton, not a blank page
  • Enforce the PBAN format (Purpose → Background → Action → Next Steps)
  • Embed rubric criteria directly into the prompt so quality expectations were visible during drafting, not revealed at review

Result: content production time dropped from 3–5 hours to under 30 minutes for standard sprint reports.

02

Sprint Post Review Report (SPRR) System

The end-of-sprint review was redesigned from an unstructured debrief into a structured artifact. The SPRR required teams to document: what was planned vs. delivered, variance causes, decision changes, and what would be different next sprint.

Not a compliance form — a decision-readiness tool. Program Managers could read one SPRR and understand where a team stood without a meeting. The meeting cost moved into the document, which doesn't require scheduling.

03

PM Verbal Scoring Check-In System

A structured PM verbal scoring check-in replaced back-and-forth async clarification chains that were inflating cycle times. Program Managers used a defined rubric to rate report quality on specific dimensions — clarity, completeness, decision-readiness, and reflection depth. This created a feedback loop specific enough to actually improve future submissions, rather than generic "please improve" comments that changed nothing.


The AI Design Rationale

Prompts as cognitive infrastructure — augmentation, not automation.

The goal was never to have AI write the reports. The goal was to remove the cognitive friction of starting — the blank page problem causing procrastination, underwriting, and structural inconsistency. Each prompt had three layers: a context layer (what this report is for, who reads it), a structure layer (mandatory sections and their purpose), and a quality layer (what distinguishes useful from compliant).

Avahana Integration

The prompt system was adapted to work within DeepThought's existing Avahana platform — not built as a standalone tool. Adoption friction was near-zero: people used the same interface they already used, with a better cognitive starting point embedded in it.


Communication Design

PBAN: the structural standard that made everything else work.

The PBAN format (Purpose → Background → Action → Next Steps) was the single highest-leverage change in the system. When every document follows the same structure, readers know exactly where to find information. Review time drops. Ambiguity drops. Decision speed increases.

I also designed the Inbox Architect workshop — a structured learning activity training participants on PBAN email writing, escalation logic, synchronous vs. asynchronous communication decisions, and meeting documentation. This converted the PBAN standard from a policy people knew about into a skill people actually had.


Outcomes

The numbers — and what they mean.

40% improvement in submission efficiency

Measured through internal tracking across sprint cycles. The combination of AI-assisted prompts, clearer format expectations, and the PM scoring check-in reduced total effort for writers and reviewers simultaneously.

Report cycle time halved: 6–8 days → 3–4 days

The primary driver: removing back-and-forth clarification chains. When reports were structured correctly from the first submission, the review cycle became a verification step, not a correction process.

3–6 hours per team per week recovered

Across 7 labs, SOP standardization and the prompt library removed the ambiguity-tax costing teams hours of rework, re-explanation, and reformatting every sprint cycle.

Decision-readiness improved for Program Managers

The shift from noise to signal meant PM check-ins moved faster, escalations were clearer, and leadership had better data for resourcing and prioritization decisions across the organization.


Reflection

"The best systems infrastructure is the kind people stop noticing — not because it stopped working, but because it started working so reliably that it became the floor, not the ceiling. That's what I was building: not a tool people had to remember to use, but a structure they couldn't help but move inside of."

Let's work
together.

Available for remote roles and fractional engagements with async-first, results-driven organizations. If your team cares about how people experience their work — not just what they produce — we're probably a good fit.

Currently available

Open to full-time remote roles and fractional engagements in People Operations, Organizational Psychology, AI Systems Design, and Behavioral Research. Async-first preferred. Mon–Fri, 6–8 hrs/day.