Ebuka – Python, LangChain, OpenAI

An AI-powered knowledge platform that captures and preserves work knowledge, recording workflow context (screen and audio), summarizing it, linking related information, and turning it into a visual, searchable memory engine to prevent knowledge loss and improve performance.

• Built and maintained dashboards for the Test Automation and Tools team;
• Documented software development processes and procedures for future use;
• Resolved complex issues related to software applications quickly and effectively;
• Collaborated with team members to ensure successful project completion within timeline constraints;
• Integrated third-party APIs into existing systems;
• Performed unit testing on application components prior to release.

An AI-powered analytics assistant for search marketers that connects directly to ad platforms like Google Ads, turning raw campaign data into plain-language insights so users can ask questions like “Why did my ROAS drop?” and get clear explanations with optimization steps in minutes, saving time and enabling faster, data-driven decisions without needing a data science background.

• Designed and implemented LLM-powered analytics agents that translate natural-language questions into structured queries over advertising data (e.g., Google Ads), enabling non-technical users to retrieve insights without dashboards;
• Built end-to-end data pipelines for ingesting, cleaning, and aggregating high-volume marketing performance data, supporting real-time and historical analysis for campaign diagnostics and optimization;
• Developed explainability layers for AI-generated insights, combining statistical analysis with language models to surface root causes behind metric changes (e.g., ROAS drops, CPC spikes) in clear, human-readable form;
• Integrated vector search and semantic retrieval to provide context-aware responses by linking current campaign performance with historical trends, benchmarks, and prior optimization outcomes;
• Collaborated with product and growth teams to iteratively refine AI outputs, improve prompt engineering, and deploy production-ready models with a focus on accuracy, latency, and user trust.

An AI-powered conversational fintech platform that lets users manage money, send and receive funds, and pay bills directly through chat on WhatsApp using natural language, with secure KYC, payments, and personalised spending insights to support better financial habits.

• Built a production-ready conversational AI platform on WhatsApp for payments, transfers, and bill automation;
• Engineered secure AI-driven data collection workflows, enforcing KYC, encryption, and policy controls;
• Integrated LLMs with backend services and business logic to orchestrate end-to-end financial transactions;
• Designed scalable serverless AI infrastructure across cloud services to handle high-volume messaging workloads;
• Applied AI safety, privacy, and compliance best practices to protect user data and ensure responsible model behaviour.

An AI-powered assessment platform that streamlines and secures online testing for educational institutions and training organisations with in-browser AI proctoring, intelligent question generation, automated grading with rubrics, and support for 11+ question types. The platform features an intelligent question-generation engine (Owl Assist), AI-based automated grading with rubric support, and fully integrated proctoring using computer vision and behavioral analysis, all accessible in-browser without the need for third-party tools.

• Designed and deployed an AI-assisted question generation engine using NLP and LLMs to auto-create questions and rubrics, reducing educator workload and improving assessment quality;
• Built a real-time in-browser AI proctoring system using computer vision and behavioural analytics to detect malpractice and securely scale online exams without extra software;
• Architected a dynamic assessment engine with 11+ question types, configurable scoring rubrics, and automated grading of subjective answers via LLM-based classifiers;
• Led AI development for a B2C study app that turns study materials into interactive, gamified quizzes with leaderboards and personalised progress tracking;
• Developed an advanced reporting layer that surfaces behavioural insights and real-time engagement analytics for educators and institutions.