Abhishek Jayant

● Re-architected the demand forecasting pipeline, eliminating data inconsistencies and enforcing industry-standard ML practices, delivering a 17% improvement in forecast accuracy and better pricing decisions. ● Developed an embedding-based product similarity framework using product descriptions to address the coldstart problem for new and low-history items, increasing recommendation accuracy by 40% and improving user engagement and cross-sell opportunities. ● Drove €5M+ inventory cost reduction by leading backtesting of inventory strategies for one of Europe’s largest pharmaceutical manufacturers, identifying gaps in legacy policies and validating Peak (UiPath) inventory AI through a POC that reduced overstock by 3% while maintaining a 98% service level. ● Strengthened enterprise-scale pricing optimisation systems for the world’s largest sports equipment manufacturer by improving model guardrails and delivering more stable, decision-ready price recommendations

● Led the development and deployment of deep learning models to accurately forecast solar energy generation from photovoltaic systems. ● Leveraged RAG with LLMs to automate the extraction of carbon emissions data from building documents, reducing manual processing time by 80% and saving clients an average of 5+ hours per building. ● Trained and optimised Azure vision models to extract key information from energy invoices, resulting in a 90% accuracy rate and saving over 100 hours of manual effort per month. ● Collaborated with a multidisciplinary team to develop an MVP analysing daily energy needs, providing personalised recommendations for low carbon technologies, and reducing clients' carbon footprint by 45%. ● Engineered appliance simulations for battery and thermal storage systems, designing algorithms to optimise renewable energy utilization, maximize carbon reduction and energy savings and minimising gas and electricity dependency in buildings.

● Architected, programmed and deployed a POC leveraging COVID-19 patient data within a knowledge graph, employing graph data science algorithms to derive insights on vaccine adverse effects, leading to acquisition of 3 new clients for the product. ● Employed clustering and regression algorithms to analyse customer behaviour trends and strategically target clients for specialised marketing, accelerating brand engagement. ● Engineered an end-to-end talent recommendation system, extracting employee data via SQL and AWS cloud using graph data science for predictions and subsequently deploying the system on cloud through CI/CD pipelines, reducing deployment time by 30% ● Employed LLM models (GPT2 and Pegasus) for precise text classification and summarization demonstrating improvements over traditional NLP approaches.

● Designed intuitive dashboards with Tableau, translated intricate datasets into compelling visualisations, and delivered impactful reporting tailored for non-technical management. ● Leveraged Python to extract, transform, and analyse large-scale datasets from Amazon S3, uncovering key performance indicators and actionable insights to drive strategic decision-making for stakeholders

● Designed a GUI-based prototype of a comprehensive crew management system leveraging industry-standard design principles and tools with Python. ● Engineered the architecture of a user management system with Django and REST APIs, delivering a scalable solution that enhanced user experience and streamlined account operations.

Modules: Data Analysis and Statistics, Data Visualisation, Data Mining, Database Design, Business Intelligence, Artificial Intelligence, Big Data Analytics and Data Ethics. Final Dissertation Mark: 80