Context Modern manufacturing organizations face increasing volatility: changing demand, supplier delays, capacity constraints, and complex multi-level BOMs make it extremely difficult to plan supply and production in a reliable and agile way. Traditional planning systems often fail to capture real-world uncertainty, resulting in late orders, shortages, and firefighting on the shop floor.

Pelico is building an AI-driven platform that bridges the gap between planning and execution , enabling teams to anticipate problems earlier, simulate scenarios, and orchestrate recovery actions.

Missions Part 1 - Optimization of stock flow and pegging strategies (in a deterministic setting) The first part focuses on improving how available supply is allocated to downstream needs, using optimized pegging and allocation strategies. Today, Pelico relies on a deterministic allocation rule (FIFO); this internship aims to evaluate alternative allocation policies that maximize service level & revenue while minimizing WIP, premature builds, and excess inventory.

You will explore different objective functions and constraints, such as:

• maximizing revenue, order coverage and OTIF
• minimizing inventory or early builds
• integrating safety constraints

Part 2 - Risk assessment and uncertainty modeling from historical data The second part studies how uncertainty can be formally quantified to support more robust decision-making in the future. The goal is to build a risk assessment layer that characterizes plant parts, supply events, or pegging links based on historical behavior, considering volatility in supply lead times and quantities, demand variability, production cycle time uncertainty, and historical shortage patterns.

This Includes

• collecting and structuring historical data
• building indicators
• estimating shortage or lateness probabilities
• propagating uncertainty through the pegging graph,
• evaluating whether current safety stocks or safety times are adequate.

In essence, this part creates the risk intelligence needed to later enhance the optimized policies explored in part 1.

During This Internship, You Will

• Conduct a literature review and benchmark of state-of-the-art methods.
• Formalize the problem, propose modeling assumptions, and design candidate algorithms.
• Build and iterate on a prototype.
• Evaluate your approach on real-world industrial datasets from Pelico customers.
• Collaborate with PMs and engineers to challenge assumptions and refine the modelling direction based on your results.
• Document your work clearly and share learnings with the team.

Expected Outcomes

• A working algorithmic prototype demonstrating the proposed approach.
• A quantitative evaluation comparing different methods and highlighting trade-offs.
• Clear recommendations for productization, including limitations and assumptions.

Success means producing a method that is sound, explainable, and applicable at industrial scale, strong enough to influence Pelico’s roadmap or be integrated into our platform.

Depending on performance and interest, the internship may lead to a full-time offer.

What You Will Learn

• How to apply advanced optimization and stochastic methods to real industrial planning & orchestration problems.
• How to turn research into impact: from algorithm design to prototyping, testing and integration into a production-grade SaaS platform.
• What it means to work in a fast-growing SaaS scale-up building the next generation of solutions bridging the gap between planning and execution.

What we offer 💡

• Work on a highly impactful product that users love!
• Office location at the heart of Paris (75002)
• Remote flexibility while autonomous
• 1 paid leave/month worked
• Any additonal days related to graduation (presentation, oral, rattrapage)
• 50% meal allowance: 10€/day worked (Swile card)
• 100% public transportation or equivalent in sustainable mobility package
• Afterwork every quarter