Applied AI & MLOps — I build production-ready AI systems and backend services.
Focused on LLM applications, API design, and deployment.

• Build production-grade LLM applications and inference services.
• Design and ship scalable backend APIs for ML systems.
• Deploy and maintain ML infrastructure (model serving, monitoring, CI/CD).
• Improve model reliability, latency, and cost through engineering and tooling.

• Languages: Python, TypeScript, JavaScript, SQL
• ML frameworks: PyTorch, TensorFlow, JAX (Deep Learning, CV, Multimodal)
• Computer Vision: SegFormer (B5), ConvNeXt, OpenCV, Semantic Segmentation, Object/Blob Extraction
• LLM tooling: LangChain, LlamaIndex, Hugging Face (Transformers, Datasets, Diffusers), RAG pipelines, OpenAI-compatible APIs)
• Backend & APIs: FastAPI, RESTful APIs, Async Python, Webhooks
• Serving & infra: Docker, Kubernetes, Terraform
• MLOps: MLflow, Weights & Biases, Model Versioning, Experiment Tracking, CI/CD for ML
• Cloud: AWS (Lambda, ECS, EC2, S3, SageMaker), GCP (Compute Engine, Cloud Storage)
• Data & storage: PostgreSQL, S3
• Frontend stack: Next.js (React), Tailwind CSS, Supabase (Auth, Storage, PostgreSQL)

• EcoVision — UAV-based plant species identification & dominance analysis system

• Short pitch: End-to-end computer vision pipeline for ecological monitoring using drone imagery.

• Tech: PyTorch, SegFormer B5, ConvNeXt, OpenCV, FastAPI, Python

• Highlights / Metrics: Semantic segmentation + blob extraction + species classification pipeline ~95% classification accuracy on trained species Automated 2×2m patch-based dominance scoring for ecological surveys Designed for research publication and field deployment

• Link: (Private / available on request)

• CallsAid — AI-powered voice & workflow automation platform for businesses

• Short pitch: Production-grade Voice AI system for automating business calls and workflows in African markets.

• Tech: FastAPI, Python, Next.js, Supabase (Auth, Storage, Postgres), OpenAI-compatible APIs

• Highlights / Metrics: Voice AI agents integrated with business workflows Role-based access (Business Owner / Department Manager) Built for multilingual expansion (Africa-first design)

• Link: https://callsaid.com

• Production-first: I design AI systems with deployment in mind, clear service boundaries, measurable latency, cost-aware inference, and failure modes considered upfront.
• Iterative & metric-driven: I run small, controlled experiments (models, prompts, pipelines), track outcomes with explicit metrics (accuracy, error rates, throughput), and keep rollback paths simple.
• Research-to-production execution: I translate research outputs (CV models, LLM workflows) into usable APIs and products without losing scientific rigor.
• Collaboration-ready engineering: I prioritize clean interfaces, readable code, and concise documentation so ML, backend, and product teams can move independently.
• Operational observability: I instrument pipelines and services with logging, metrics, and alerts to catch model drift, regressions, and data issues early.

• Applied AI and MLOps roles (cofounding, consulting, contracting, full-time)
• Collaborations on LLM products and scalable inference systems
• Speaking or writing about production ML engineering and model ops

• GitHub: github.com/Chibu4ril
• Website: -
• Email: chibuokemonye@gmail.com
• LinkedIn

• I focus on bridging research into reliable, production-grade AI systems.
• I care deeply about reproducibility, cost-efficient model serving, and developer ergonomics.
• I build at the intersection of AI research, full-stack engineering, and product strategy.
• Outside of work: I study philosophical wisdom, systems thinking, and long-term human progress.

Made with ❤️