Inessa Duc

Combining advanced analytics with applied AI across real-world projects in natural language processing, agentic systems, and predictive modelling. - Engineered a web application using Wikipedia as the data source that generates structured industry reports from a single industry name input, achieving 90% accuracy through applied guardrails. - Designed an LLM-powered agent integrating five data sources via APIs, including real-time geopolitical news and trade flows, to help businesses assess sourcing risk a

Joined M&S's Food Data Science team for a 4-month dissertation project as part of my MSc in Business Analytics at UCL, chosen specifically to apply my analytics background to a live, high-stakes business problem: M&S has committed to doubling food sales while cutting food waste by 50% by 2030, and my goal was to identify the main drivers behind that waste, so the business could have more actionable interventions. - Developed predictive machine learning models using large-scale product and store data, narro

Ipiranga is Brazil's largest fuel distribution network, operating over 6,500 service stations. Guided executives in defining and tracking performance targets and managed the governance of the variable compensation model. Led a team of 3 analysts. - Redesigned the corporate variable compensation model by analysing more than 15,000 individual performance goals, identifying misalignment with executive priorities and increasing alignment from 50% to 80%.

Worked with corporate teams to improve how performance was defined, measured and managed, helping leaders turn strategic priorities into clear goals and execution focus. - Proactively redefined an initial dashboard request by guiding managers on metric relevance and maturity, then formulating 40 KPIs to support IT strategic goals.

Rede D'Or is Latin America's largest private healthcare network, operating more than 75 hospitals and 60 oncology clinics across Brazil. Led performance improvement initiatives in the hospital billing chain, working across clinical, commercial, operations and technology teams to drive data adoption and operational efficiency. - Designed and delivered a three-day performance management workshop for 100+ nurses, combining internal best practices, live dashboard training, and external expertise in change man

Americanas is one of Brazil's largest retail groups, with over 1,600 stores across all states and a major e-commerce platform. Led demand planning and purchasing decisions using sales, inventory and supplier data to balance availability and capital allocation. Oversaw monthly purchase for more than 5,000 SKUs. Led a team of 9 analysts. - Initiated and led a Lean Six Sigma project with cross-functional teams, applying statistical analysis and root cause techniques to raise supplier fill rates, reducing gap

Managed sales forecasting and purchasing planning, negotiating directly with suppliers to optimise product assortment and drive revenue growth. - Conducted Market Intelligence study to identify regional pricing opportunities by city/state, increasing margins in electronics sales.

Selected for a highly competitive graduate programme (over 15,000 applicants), rotating across operations, distribution and commercial teams to gain broad exposure to retail decision-making. - Mapped 100% of the activities of 170+ employees, prioritising and scoping 3 automation projects with potential savings of approx. 8,500 working hours per year.

MRS Logística operates one of Brazil's largest rail networks, connecting 107 municipalities and serving the country's highest GDP-generating regions. Managed preventive maintenance planning and monitored equipment availability and productivity across the rail network. - Improved production reporting across 200 railway assets by mobilising and training operational teams on accurate data entry, reducing input errors by 95%.

Learn more about the course here: https://www.ucl.ac.uk/study/prospective-students/graduate/courses/business-analytics-msc - Engineered a web application using Wikipedia as the data source that generates structured industry reports from a single industry name input, achieving 90% accuracy through