Revenue quality, decomposed
Took a SaaS company’s revenue-quality claims apart using cohort analysis, survivor-weighted ARPA inflation and expansion-type decomposition — to test whether reported NRR reflected durable expansion or a churn artefact.
I build AI that
moves money, wins
attention, and proves
the numbers.
Founder and technology operator across African fintech, AI advertising and consumer markets. I built the AI infrastructure behind a cross-border payments platform serving 10M+ users — solo — and now build Fakt, provenance-first diligence software for Africa-focused investors.
0
Users reached
Kredete consumer + Hinstant B2B
0
Ventures built
Founded, co-founded or led technology
0
AI hire at Kredete
Designed the whole stack solo
2021
Shipping since
Nigeria · USA · Paris
Position
I have built things on both sides of the table — the engineering that makes a product work, and the operating judgment that makes it worth funding. Emerging-market businesses fail on verification, unit economics and distribution long before they fail on technology. I work where those meet.
01 — Selected work
Things I’ve built
Things I’ve built
and run.
-
01
Fakt
Founder · 2026 – Present · AI diligence for emerging-market investors
Provenance-first diligence software for Africa-focused funds and DFI-adjacent investors. A company’s data room goes in — deck, financial model, cap table — and every metric comes out cited to its exact source, cross-checked across documents, with conflicts and unverifiable figures flagged. Built around the real work of an investment analyst: thesis-fit screening, and a data-room-to-draft-memo pipeline. Distant capital under-allocates to African deals because verification is hard; Fakt is a direct attack on that.
-
02
Mixtape — SocialGeneration
Chief Technology Officer · 2024 – Present · Atlanta, USA (remote)
I lead technology for a US-based AI advertising company and am the architect of Mixtape, its core platform — built on a proprietary ad-classification framework and performance-scoring system. The role runs across both sides of the business: product architecture and engineering delivery on one, commercial trade-offs on infrastructure cost and platform direction on the other.
-
03
Kredete & Hinstant
Founding AI Engineer · 2023 – 2026 · 10M+ users
First AI hire. I designed and built, solo, the AI systems powering the consumer product — cross-border transfers, remittance, FX, payments, agentic commerce — and the B2B arm Hinstant, covering automated business payments, cards and remittance. Every build decision was governed by unit economics: workloads routed across model providers by cost and latency, smaller open-source models fine-tuned where general-purpose ones were too expensive per transaction.
-
04
Stocked
Co-Founder · 2024 – 2026 · Multi-city, Nigeria
A fresh grocery delivery business across multiple Nigerian cities, built on WhatsApp and voice-AI ordering infrastructure adapted to local language and buying behaviour. I developed the embedded-finance roadmap — credit and payments as the structural margin layer — and researched the full Nigerian licensing ladder, from state money-lender permits through CBN licensing, alongside the DFI capital structures to fund it.
02 — The engineering layer
Under the hood
Under the hood
of a 10M-user
payments stack.
Fintech outputs carry legal and financial consequences. Nothing here shipped on a vibe-check — this is how the Kredete AI infrastructure was actually assembled, and what I bring to the next one.
-
P.01
Orchestration
Built a multi-agent orchestration layer on AWS Bedrock with Strands Agents, then migrated it to LangChain running on Cerebras once customer-facing inference latency became the binding constraint.
-
P.02
Model routing
Multi-provider routing across OpenAI, Anthropic and Google Gemini, selected per task on latency and cost profile rather than defaulting to one vendor. Cost per transaction was a first-class design input.
-
P.03
Retrieval
RAG on pgvector in production, grounded in policy documents, live transaction context and product knowledge — so agents reason against real ground truth instead of training-data recall.
-
P.04
Fine-tuning
Fine-tuned open-source Hugging Face models with LoRA and QLoRA adapters for fintech-specific tasks where general-purpose models were too expensive or too imprecise. Validated generalization with gradient and learning-curve analysis.
-
P.05
Evaluation
Evals from day one — a fixed task corpus and a regression harness, so every prompt and model change was measured before it shipped. This is the part most teams add after their first incident.
-
P.06
Product surface
Before the AI work: the Angular SPA behind Kredete’s consumer product — dashboards, transaction flows, KYC, account management, dynamic Reactive Forms validation over sensitive financial data, and REST integration of transfer, balance and FX-quoting APIs.
Languages
Python (primary) · TypeScript · JavaScript · SQL
LLM providers
OpenAI · Anthropic · Google Gemini · Cerebras · Hugging Face
Agents & orchestration
LangChain · LangGraph · AWS Bedrock · Strands Agents · RAG pipelines · autonomous workflows
Vector databases
pgvector in production; architecturally fluent in Pinecone, Weaviate, ChromaDB, FAISS
Fine-tuning
LoRA · QLoRA · gradient & learning-curve analysis · model architecture analysis
Backend & data
FastAPI · Flask · REST design · PostgreSQL + pgvector · MongoDB
Cloud & infra
AWS (Bedrock, S3) · Firebase · Heroku · Docker
Domain
Fintech · payments · remittance · FX · agentic commerce · B2B SaaS
03 — Investment & analytical work
Reading a business
Reading a business
the way an investor
has to.
Venture & PE metric fluency
Working fluency in MOIC, IRR, NRR mechanics, unit economics and cohort behaviour — applied to live pitch materials and financial models, not to textbook cases.
African market mapping
Mapped the investment landscape across East, Southern, Francophone and North Africa: fund strategies, DFI capital structures, and the regulatory environments governing financial services.
Technical due diligence
Able to interrogate a technology company’s stack, its real infrastructure costs and its claims directly — because I have priced, built and run the same systems.
04 — Path
Five years,
Five years,
three continents.
- 2026 — Present
Founder
Fakt — AI diligence platform for emerging-market investors
- 2024 — Present
Chief Technology Officer
SocialGeneration — AI advertising, Atlanta, USA (remote)
- 2024 — 2026
Co-Founder
Stocked — fresh grocery delivery, multi-city Nigeria
- 2024 — 2026
Founding AI Engineer
Kredete — cross-border payments fintech, 10M+ users (remote)
- 2023 — 2024
Frontend / Angular Engineer
Kredete — consumer web application, KYC and transaction flows
- 2022 — Present
Web Developer
GrowthPhysics, USA (remote) — marketing sites and web applications
- 2021 — 2022
Web Developer
Flywallet, Paris (remote) — travel & tourism
- 2024
B.Eng, Electronic Engineering
University of Nigeria, Nsukka
Let’s build
something.
amehotuoche@gmail.com
Phone+234 815 794 3265
Based
Port Harcourt, Nigeria
GitHubgithub.com/otuoche ↗
LinkedInfaithfulness-ameh ↗
DocumentDownload CV ↗