Microscopic Imaging for Hematology Analyzers

Nvidia Delivery ArmSemantic Search, RAG and GenAI* Responsible for ideation, solution architecture and day-to-day technical guidance and execution of multiple NLP and LLM-based projects* Search and RAG on technical documents and structured databases* Modeling of a large database of natural conversations as a graph using GenAI to derive business insights * Classification of Linux command severity for cybersecurity surveillance of a large bank* Vision transformer based approach for identification of chemical molecule from hand-drawn diagrams* Delivered sessions on using Nvidia's frameworks for GenAI to multiple organisations* Delivered webinars and co-wrote technical blogs on improving speech-based conversational agents* Supporting pre-sales and marketing with ideation, feasibility analyses and demos

Helping to build and scale a Job matching system for "Opportunity Youth" in Mombasa.

AI CoE, Medical & Life Sciences * Servant-led a team of 4 ML engineers on projects in surgical analytics (bleeding detection and workflow analysis) * Helped build "Chest-rAI" - An AI-based chest X-Ray analysis system to find, localize and report abnormalities, which has won the "Big Innovation Award USA - 2021" and ASSOCHAM's "2nd Innovators' Excellence Award" * Co-inventor in 5 patents in medical imaging and NLP * Actively publishing academic and whitepapers * Mentoring interns * Pre-sales and planning (writing proposals, task tracking, risk management) * Interfacing with IT and platform teams for GitHub migration and infrastructure issuesChest-rAI:> Auto-encoder and clustering for unsupervised data cleaning> Classification and localisation of abnormalities on chest X-Rays> Label extraction, summarization, structuring and generation of X-Ray reports> Development of anonymization tools and annotation tools; management of annotation teams> Development of a multi-model inference pipeline and deployment on cloud (kubernetes)> Deployment monitoring and automation with Terraform and GitHub Actions> Inference optimization on Intel and Nvidia hardwareSurgical analytics> Exploration and planning for polyp detection in colonoscopy (software defined endoscopy using Nvidia's Clara Holoscan)> Generation of synthetic data with blood-flow simulation, detection and quantification of blood in laparoscopic surgery videos> Deep-learning based analysis of surgical workflow in laparoscopy videos

Digital Image Processing Team* Deep learning for computer-aided-diagnosis in radiology.* Assessed clinical relevance and explored issues of explainability, generalizability and fairness.* Replicated previous work and experimented with several training refinements.* Achieved competitive results for TB detection.