Data Scientist, Liquidity Products Squad

Job description

**What is Cobre, and what do we do?** Cobre is Latin America’s leading instant b2b payments platform. We solve the region’s most complex money movement challenges by building advanced financial infrastructure that enables companies to move money faster, safer, and more efficiently. We enable instant business payments—local or international, direct or via API—all from a single platform. Built for fintechs, PSPs, banks, and finance teams that demand speed, control, and efficiency. From real-time payments to automated treasury, we turn complex financial processes into simple experiences. Cobre is the first platform in Colombia to enable companies to pay both banked and unbanked beneficiaries within the same payment cycle and through a single interface. We are building the enterprise payments infrastructure of Latin America! **What we are looking for:** We're looking for a Data Scientist to build the quantitative engine behind Cobre's Liquidity Intelligence Platform—the models that forecast cash flow across every account and rail, decide how to fund positions ahead of need, and price liquidity and FX risk in real time. This isn't a role where you hand a notebook to someone else to productionize. You'll build forecasting and optimization models, ship them into a live production stack, and own the validation methodology that proves they work. You'll also have a technical dotted line into Cobre's Data & AI Center of Excellence, giving you a forum to stress-test your methodology with peers working on similar problems elsewhere in the company. The squad's philosophy: build the industry's leading-edge real-time liquidity management system—one that enables exponential scaling of payment volumes through industry-leading liquidity efficiency, starting with a real-time monitoring and decisioning engine and building toward autonomous execution. **What You'll Own:** *Cash Flow Forecasting* Build and maintain the forecasting models that predict inflows and outflows across every bank account, rail, and currency (COP, MXN, USD, USDT/USDC)—starting from payout volume forecasting for banking suppliers and extending across the full liquidity network. You'll own model selection, feature engineering, backtesting discipline, and the retraining cadence that keeps forecasts sharp as volume scales. Forecast quality is the ceiling on everything downstream—funding recommendations, risk classification, and capital efficiency all inherit the forecast's error. *Liquidity Optimization & Decisioning* Design and maintain the optimization models (e.g. linear/mixed-integer programming over time-expanded network structures) that turn a forecast into a funding recommendation: how much to move, from where, to where, by when, at what cost and risk. You'll define the constraint set, tune the objective function, and work with engineering to get the solver's output into a decision an operator can act on in seconds. **What would you be doing:** - Build production models — own the full lifecycle: problem framing, feature engineering, model selection, backtesting, deployment into the production stack, and monitoring. - Build with an AI-native mindset — use AI tools to speed up your own modeling loop—generating candidate features, scaffolding backtests, querying data, or prototyping a solver formulation—so you spend your time on judgment calls, not boilerplate. You bring working proofs of concept into squad discussions, not just ideas. - Partner with the PM and engineers day-to-day — you're not handed a spec and asked to model it in isolation; you help define what's worth modeling.. - Own validation and trust — design the backtests, baselines, and live-monitoring metrics that prove a model works before it gets more autonomy, and that catch drift before it causes a loss. - Work close to the money — your models directly touch real financial exposure (funding decisions, NSF/overdraft risk, FX pricing)—there's no abstraction layer between your model output and a dollar moving. - Translate model behavior for non-technical stakeholders — explain what a model is doing and why in terms Treasury, Finance, and Sales can act on, and translate their constraints (covenants, risk appetite, corridor economics) back into modeling choices. - Stay close to the data — work directly in data and AI stack; you should be comfortable enough with the data to sanity-check your own model's inputs and outputs without waiting on a data engineer. **What do you need:** - *4+ years in a data science, quantitative, or applied ML role, ideally in fintech, payments, banking, or another domain with real financial exposure* - Education: Bachelor's or Master's in Statistics, Applied Mathematics, Operations Research, Physics, Econometrics, Computer Science, or a related quantitative field. A Ph.D. is a plus but not required. - Experience with optimization techniques—linear programming, mixed-integer programming, or graph-based network optimization—applied to a real operational problem - Comfortable designing and running rigorous validation: backtesting, baseline comparisons, A/B or online experimentation for models that make decisions with real consequences. - Strong Python and SQL skills; experience owning a model from raw data through to a production pipeline. - Strong analytical foundation—comfortable communicating uncertainty, trade-offs, and model limitations to non-technical stakeholders. *Product & Technical Acumen* - Ability to translate a business or treasury problem into a well-posed modeling problem, and translate model output back into a clear, actionable recommendation - Comfortable with data pipelines and real-time or near-real-time systems, not just batch analysis - Able to read and contribute to technical architecture discussions with engineers, and to defend modeling choices under scrutiny - Exceptional written communication—clear model documentation and reasoning, not just code - Comfort with ambiguity—you'll help define which problems are worth modeling, not just execute a fixed research agenda *Nice to Have* - Hands-on experience with Snowflake (dynamic tables, tasks, Streamlit) and modern Python ML/optimization libraries (e.g. LightGBM, PuLP or similar solvers) - Experience with multi-armed bandit or reinforcement-learning approaches to pricing or resource allocation. - Background in treasury management, cash forecasting, or FX/corridor pricing - Exposure to stablecoins or crypto rails (USDT/USDC, Ethereum, Tron, Solana) - Previous startup or scale-up experience (Series A/B stage)