Lawson Ekhorutomwen

Full-stack developer and AI engineer who builds from first principles.

Production web and mobile apps alongside AI systems, backpropagation, retrieval, computer vision, implemented from scratch, not assembled from APIs. Frontend Engineer at Sendpiper, studying Medicine and Surgery at the University of Benin.

View projectsDownload resume
Lawson Ekhorutomwen

About

Full-stack developer and AI engineer, and a final-year medical student. Built and shipped production web and mobile applications alongside AI systems built from first principles, not course certificates: implemented backpropagation, attention, and a GPT model from scratch, reproduced a peer-reviewed clinical AI paper, built a clinical NLP pipeline evaluated against ground truth I hand-annotated myself, and built a multi-agent clinical scribe that verifies its own AI-generated notes before a doctor sees them. Long-term focus: AI applied to medicine.

Medicine and SurgeryUniversity of Benin (UNIBEN) · Benin City, Edo StateExpected 02/2027

Projects

01/09
Doctor's hand writing in a patient's medical chartAI & Machine Learning

AI Clinical Scribe

QA recall: single-pass inconsistent → 3-pass reliable

A multi-agent pipeline that turns a doctor-patient consult recording into a reviewed, editable SOAP note, with every AI claim checked against the transcript before a doctor sees it.

Whisper transcription and pyannote.audio speaker diarization feed a LangGraph extraction agent, which is checked by a separate QA agent and validator agent before a doctor reviews and finalizes the note. Controlled testing with deliberately planted errors showed a single QA pass caught real issues inconsistently across identical runs; running three passes and combining the flags measurably improved recall. FastAPI backend, PostgreSQL with a full field-level edit history separate from the original AI output, React/TypeScript/Tailwind review interface with live mic recording.

LangGraphLangChainGroqWhisperpyannote.audioFastAPIPostgreSQLReactTypeScriptTailwind CSS
Source code

Experience

Frontend Engineer · Sendpiper

Jul 2025 – Present

Meta-certified messaging automation platform for WhatsApp, Instagram & Messenger

  • Built and maintain the customer-facing web application in React and Next.js, including a drag-and-drop flow builder for automated multi-channel messaging campaigns and a unified inbox for managing WhatsApp, Instagram, and Messenger conversations in one interface.
  • Built the contact/audience management and campaign analytics interfaces, giving users visibility into a broadcast pipeline processing 200,000+ messages per day at 99.9% reliability.

Skills

AI & ML Engineering
Neural Networks (Built from Scratch)Backpropagation & Autograd EnginesTransformer Architecture & AttentionLLM Training (GPT)LoRA Fine-TuningRetrieval-Augmented Generation (RAG)Multi-Agent Systems & LLM Orchestration (LangGraph, LangChain, Groq)Speech-to-Text & Speaker Diarization (Whisper, pyannote.audio)Explainable AI (SHAP, Grad-CAM, Integrated Gradients)Clinical NLP & Named Entity RecognitionComputer Vision (MediaPipe, OpenCV)MLOps (Model Serving & Monitoring)PyTorch
Languages
PythonTypeScript / JavaScript
Frontend
ReactNext.jsVue.jsReact Native / Expo
Backend
FastAPIFlaskNode.js / ExpressSQLAlchemy
Infrastructure
AWSGoogle Cloud Platform (GCP)PostgreSQLDocker

Get in touch

Open to engineering roles, especially where medicine and AI meet. Reach out directly or download the resume below.

lawsonctrl@gmail.comGitHubLinkedInResume (PDF)

Benin City, Edo State, Nigeria