Lead Operations Researcher

About Praxas AI

Praxas AI is the AI-native decision layer for supply chains. We sit on top of the ERP a company already runs and continuously re-optimize what to make, stock, buy and ship as conditions change, helping four primary departments: sourcing, inventory, production planning and logistics.

The Opportunity

We're hiring our first Operations Research Engineer to own the math behind every Praxas decision. This is a hands-on build role. You'll build the optimization engine for production planning first, then inventory, sourcing and logistics. You will turn messy factory realities into models that run on real customer data and re-plan continuously. You will work directly with the founders, our AI and data engineers, and the planners who use what you build every day.

What You Will Do

• Build the decision engine
• Model real plants: capacities, routings, bills of materials, changeovers, labor, material availability and due dates.
• Formulate and solve production planning and scheduling problems (lot sizing, sequencing, capacity allocation) that minimize lost sales and cost.
• Pick the right method for each problem: mixed-integer programming, constraint programming, heuristics or decomposition.
• Keep solve times fast enough to re-plan on demand for plants with hundreds of products.
• Design models that are configured per customer (industry, production type, demand pattern), not rewritten per customer.
• Make re-optimization continuous and trustworthy
• Build rolling-horizon re-planning that reacts to new orders, delays and breakdowns without reshuffling the whole plan.
• Use time fences, frozen windows and change penalties so planners can trust the schedule.
• Turn the trade-off between optimality and stability into a setting planners understand.
• Model uncertainty in demand, lead times and yields with stochastic or robust approaches.
• Extend into inventory optimization and forecasting
• Build inventory optimization: safety stock, reorder points, service-level targets and multi-location policies.
• Use probabilistic forecasts (ranges, not single numbers) as optimizer inputs, working with our forecasting engine.
• Backtest every model against customer history before it touches a live plan.
• Ship with customers and the team
• Work directly with planners and plant heads at our design partners to validate constraints and results.
• Feed our AI agents the decisions they act on: delay a production run, expedite a PO, adjust safety stock.
• Tell our data engineers exactly what data each model needs, and how to handle missing or messy inputs.
• Define and track the metrics that prove value: lost sales avoided, inventory freed, on-time-in-full and machine utilization.

What Success Looks Like

• In the first 90 days
• Ship v1 of the production planning model, running on real data from our first design partners.
• Set a baseline for each partner: current lost sales, utilization, changeover time and plan versus actual.
• Agree solve-time and plan-stability targets with the team, and meet them on partner data.
• Document every model's formulation, assumptions and parameters so others can extend it.
• Within 6–12 months
• Run continuous re-optimization in production for every customer, with stability controls planners trust.
• Deliver results customers sign off on: fewer lost sales, less excess inventory and better on-time-in-full.
• Connect inventory optimization and probabilistic forecasts to production planning in one engine.
• Build reusable model templates by production type (job shop, batch, assembly, continuous) to cut implementation time.
• Own the operations research roadmap, and help hire and mentor as the team grows.

Requirements & Tech Stack

• Operations research and advanced modeling: design and implement advanced analytical models using techniques such as mixed-integer linear programming, discrete event simulation, Monte Carlo simulation, stochastic modeling and decision analysis under uncertainty.
• Experience building optimization or planning models that real operations ran on, beyond academic work.
• Strong formulation skills in mixed-integer and/or constraint programming.
• Experience with optimization solvers (Gurobi, CPLEX, OR-Tools, etc.).
• Experience with simulation tools such as AnyLogic, Simio, Arena or similar.
• Working knowledge of production planning: master scheduling, MRP, capacity planning, lot sizing and sequencing.
• Solid probability and statistics, including demand uncertainty and safety stock math.
• Production-quality Python and comfortable SQL.

Qualifications

• Advanced degree (MS or PhD preferred) in operations research, industrial engineering, applied mathematics, computer science, data science or a related quantitative field.

Preferred

• Experience at a supply chain planning software company (such as o9 Solutions, Blue Yonder, Kinaxis, SAP IBP or Coupa), or on an in-house planning or OR team at a manufacturer, retailer or 3PL.
• Rolling-horizon planning and schedule-stability methods used in production systems.
• Inventory optimization: service-level targets, (s, S) and (R, S) policies, multi-echelon networks.
• Forecasting for intermittent or lumpy demand (Croston, SBA, TSB) and probabilistic forecasts.
• Heuristics, metaheuristics or decomposition for large problem instances.
• Process industries (chemicals, food, CPG): batches, shelf life, sequence-dependent cleaning.
• Experience with Indian mid-market manufacturers or their ERPs (Tally, SAP Business One).
• APICS CPIM or CSCP certification.
• Experience pairing optimization with LLM or agent systems.
• Familiarity with Silver, Pyke and Thomas, Inventory and Production Management in Supply Chains, the foundation of our engine.