ML Systems Engineer · Agentic AI · Credit Risk & Quant Analytics Open to remote & international engagements — Freelance · Contract · Full-time
Production-grade ML engineer with 5+ years building systems that run in the real world — not just notebooks. My work spans agentic pipelines, RAG backends, forecasting infrastructure, credit risk engines, and data platforms, serving clients across fintech, telecom, SaaS, and fixed income.
Top Rated on Upwork · 100% Job Success Score
Agentic AI & RAG Systems LangGraph multi-agent pipelines (supervisor + sub-agent architectures), production RAG backends with pgvector + BM25 + cross-encoder reranking, LLM-powered analytics assistants, Manuscript Companion Chat with Voice Drift Analytics, and GenAI workflows with structured outputs. Built across Groq/Llama, OpenAI, and custom fine-tuned models. Audited LoRA fine-tunes at production scale — diagnosed a ~50pp BFCL regression on a Qwen-2.5-7B model.
ML Systems & Data Infrastructure End-to-end pipelines from raw ingestion to deployed inference. Real-time platforms with FastAPI, PostgreSQL/TimescaleDB, and Azure. Prophet and Darts forecasting. Load-tested at 30,000 subscribers — 5.4M records processed at 1,480 subscribers/second, 100% prediction success rate. Salary prediction pipelines, churn engines, and production ML assessments delivered under tight deadlines.
Credit Risk & Quant Analytics Scorecard development, PD/LGD/EAD modelling, and portfolio analytics. XGBoost/Logistic Regression credit models (15% improvement in approval precision at Renmoney). MBS spec pool pay-up analyzers and CMO inventory matching engines for fixed income clients. Full corporate credit risk frameworks for structured lending.
Data Engineering & Automation dbt, Prefect, PySpark, SFTP ingestion pipelines, Snowflake Semantic Layer + dbt MetricFlow, macroeconomic data extraction across 54 African countries. gRPC monorepo refactoring for production-scale data acquisition. I connect pipelines to decision surfaces.
Full-Stack Products Firebase multi-vendor marketplace backends, Chrome extensions, PWAs, and Android apps. I build products end-to-end when the problem demands it.
| Project | What it does |
|---|---|
deal-intelligence-agent |
8-node LangGraph pipeline — M&A, credit risk, distressed asset signals across SEC EDGAR, African exchanges, global news |
nigerian-fintech-agent |
7-layer fraud intelligence platform — agentic reasoning over transaction signals, CBN/EFCC/NFIU regulatory citations, Nigerian language routing, NDPA 2023 audit logging |
knowledge-rag-api |
Production RAG backend — pgvector + BM25 + RRF fusion, cross-encoder reranking, ACL filtering, atomic chunk publishing |
snowflake-semantic-agent |
Conversational analytics agent over Snowflake — dbt MetricFlow, Semantic Layer API, Horizon Catalog integration |
pulse-agent |
LangGraph multi-agent system built for DSN × BCT Hackathon — real-time signal orchestration across multiple sub-agents |
| Project | What it does |
|---|---|
churn-intelligence-platform |
End-to-end churn prediction platform with explainability and intervention routing |
corporate-credit-risk-framework |
Structured credit risk framework for corporate lending — PD/LGD/EAD modelling, scorecard calibration |
narrative-harm-classifier |
NLP classifier — Precision 1.000, FPR 0.000 on held-out test set |
nexus-quant-framework |
Quantitative analytics framework for financial modelling and strategy backtesting |
| Project | What it does |
|---|---|
mvno-intelligence-hub |
Azure-deployed subscriber analytics — Prophet forecasting, TimescaleDB, FastAPI, live SFTP ingestion every 15 mins |
chirp-api-refactoring |
Production audit & refactor of a Python gRPC monorepo — service decomposition, interface contracts, rapid delivery under assessment conditions |
talentpulse-salary-ml |
End-to-end ML salary prediction pipeline — feature engineering, model training, inference API |
| Project | What it does |
|---|---|
laetiva-backend-sandbox |
Firebase multi-vendor marketplace backend — 67/67 and 61/61 acceptance tests across two milestones, real client delivery |
the-scribe |
AI writing assistant for Christian authors — FastAPI + PostgreSQL + pgvector, RAG-powered Manuscript Companion Chat, Voice Drift Analytics, mobile-responsive |
scrollbreak |
Full-stack doomscrolling intervention — Chrome extension, PWA, Android app, Express backend |
Ecclesia-CheckIn |
QR-based fraud-proof attendance system — live deployment to a real church community |
Languages: Python · R · SQL · JavaScript ML/Forecasting: XGBoost · LightGBM · Prophet · Darts · Scikit-Learn · PyTorch · SHAP Agentic & AI: LangGraph · LangChain · pgvector · FAISS · OpenAI · Groq/Llama · Whisper · RAG pipelines Data Engineering: PostgreSQL · TimescaleDB · Snowflake · BigQuery · dbt · PySpark · Prefect · GitHub Actions Infrastructure: Azure · AWS · GCP · FastAPI · gRPC · Docker · Firebase · SQLAlchemy · REST APIs Visualization: Power BI · Plotly/Dash · Streamlit · Metabase · Recharts
- ML Systems Engineer (Freelance) — MVNO subscriber analytics, RAG backends, agentic pipelines, credit risk systems, quant tools, full-stack products
- Data Research Analyst, Automation & AI — Stears
- Credit Risk Analyst (Data Science) — Renmoney Microfinance Bank
- Data Scientist — Nupat Technologies
- Education: Data Analytics — Turing College (Digital Explorers II Scholar) · BSc Business Administration
Open to freelance, contract, and full-time roles (Remote · Hybrid · Onsite) — ML systems, agentic AI, credit risk modelling, and data infrastructure.