Devarshi is available for hire
Hire DevarshiDevarshi Raval
Verified Expert in Engineering
Data Science Engineer and Developer
Ahmedabad, Gujarat, India
Toptal member since August 5, 2025
Bio
Devarshi helps companies unlock hidden profits through mathematical optimization. With 8+ years of experience, he builds fast, tailored decision-support tools that streamline supply chains, improve scheduling, and boost pricing efficiency. Devarshi has worked with startups and global enterprises across eCommerce, retail, manufacturing, and consumer goods. He has saved millions by solving complex planning problems using LP/MIP models, constraint programming models, and heuristics in Python.
Portfolio
ConvexIQ Solutions
Python, PuLP, Pyomo, Gurobi, Google OR-Tools, CPLEX, XPRESS, Hexaly, Ipopt...
Jiffy.com
Python, Pyomo, Google OR-Tools, Constraint Programming, Algorithms, Heuristics...
ORMAE
Python, Pandas, NumPy, SQL, PostgreSQL, Jupyter Notebook, Cbc, Gurobi, CPLEX...
Experience
- Linear Programming - 8 years
- Operations Research - 8 years
- Constraint Programming - 8 years
- Mathematical Modeling - 8 years
- Supply Chain Optimization - 8 years
- Mixed-integer Linear Programming - 8 years
- Algorithms - 8 years
- Python - 8 years
Preferred Environment
Linux, Python, Visual Studio Code (VS Code), Git, Jupyter Notebook
The most amazing...
...model I've built slashed $2.5 million in annual supplier costs for a US-based retailer by building a custom optimization engine using MIP in Python.
Work Experience
Founder & CEO
2022 - PRESENT
ConvexIQ Solutions
- Created a make-to-order job shop scheduling solution using constraint programming for a biopharmaceutical client. Reduced makespan by 14% and increased throughput by minimizing machine idle times in a complex sequencing environment.
- Developed a store-level quadratic pricing optimization model for a global fast food chain. Enforced ladder pricing and constant upcharge constraints while maximizing net revenue at store levels.
- Engineered a quadratic programming-based offer optimization model to select top offers per customer segment. Accounted for price elasticities, offer types (weekend/all-day), and promotion saturation.
- Designed an integer programming model for annual promotional planning for a global brewery. Handled constraints on promo frequency, cannibalization, and product family and budget allocation, improving gross margin.
- Built a genetic algorithm-based flight optimization engine for a global brewery’s multi-touchpoint campaign—i.e., TV, digital, and out-of-home advertising. Maximized volume uplift with budget allocation constraints and media response curves.
- Developed a time-space network model for locomotive scheduling with maintenance and home-shed constraints. Achieved the same operational coverage using 7% fewer locomotives.
- Created a tax-harvesting MIP model tailored to Indian financial regulations to maximize long-term capital gains within exemption limits while minimizing short-term capital gains. Handled stocks, mutual funds, ETFs, and bond holdings.
- Built a Monte Carlo simulation-based mutual fund analyzer that tested SIP and lump-sum returns over randomized date windows, providing statistically robust comparison metrics and rankings.
- Developed a tail assignment mixed-integer programming (MIP) model for a commercial aviation software that reduced overall aircraft idle time and operational cost, while maintaining constraints on turnaround time and maintenance windows.
Technologies: Python, PuLP, Pyomo, Gurobi, Google OR-Tools, CPLEX, XPRESS, Hexaly, Ipopt, Column Generation, Decomposition Methods, Cbc, Linear Programming, Quadratic Programming, Nonlinear Optimization, Genetic Algorithms, Algorithms, Constraint Programming, Git, SQL, Monte Carlo Simulations, Scheduling, Taxation, Portfolio Analysis, Network Optimization, Supply Chain Optimization, Amazon EC2, AWS Lambda, Azure, Cost Reduction & Optimization (Cost-down), Operations Research, Pricing Optimization, Heuristics, Simulated Annealing, Mathematical Modeling, Prescriptive Analytics, Decision Intelligence, Statistical Analysis, Mixed-integer Linear Programming, Communication, Microsoft PowerPoint, Stakeholder Engagement, Streamlit, SimPy, Vehicle Routing, Task Scheduling, Python API, FICO Xpress, Scientific Data Analysis, AI Agents, Artificial Intelligence (AI), Back-end, Combinatorial Optimization, Time Series Forecasting, Machine Learning, Analytics, Consumer Packaged Goods (CPG), PySpark, Pricing Models, Data Science, Databricks
Operations Research Engineer
2021 - 2024
Jiffy.com
- Built a multi-objective assignment model using Google OR-Tools with CP-SAT solver to optimally assign suppliers to customer orders based on delivery time, cost, and shipment count. Reduced fulfillment cost by 1.5%.
- Designed a customer segment-level assignment strategy framework to determine ideal supplier priority mixes. Validated results through A/B testing across segments and iteratively improved fulfillment performance at lower costs.
- Created a simulation tool for customer order assignments that benchmarked supplier competitiveness based on KPIs. Performed statistical significance tests to support pricing and lead-time negotiations with suppliers.
- Developed a max-flow-based network optimization model for dynamic multi-product bundle searches. Reduced average query time from 5 seconds to just 0.12 seconds.
- Combined heuristics and constraint programming for 2D layout optimization of rectangular items. Reduced raw material wastes by 6%.
Technologies: Python, Pyomo, Google OR-Tools, Constraint Programming, Algorithms, Heuristics, Simulations, Amazon EC2, AWS Lambda, Amazon SageMaker, Pandas, NumPy, Jupyter Notebook, A/B Testing, Supply Chain Optimization, eCommerce, eCommerce Fulfillment, Operations Research, Network Optimization, Cbc, Computational Infrastructure for Operations Research (COIN-OR), Cost Reduction & Optimization (Cost-down), Mixed-integer Linear Programming, Communication, Mathematical Modeling, Decision Intelligence, SimPy, FICO Xpress, Scientific Data Analysis, Back-end, Combinatorial Optimization, Time Series Forecasting, Analytics, Consumer Packaged Goods (CPG), PySpark, Pricing Models, Data Science
Senior Operations Research Scientist
2019 - 2021
ORMAE
- Built a large-scale MIP model for a cement manufacturer to optimize multi-modal (rail and road) production-distribution planning. Reduced total logistics cost by 1.2% using Gurobi with pan-India scope.
- Solved a capacitated vehicle routing problem with time windows, pickups and deliveries, and multiple trips to minimize the longest route for a furniture eCommerce client. Reduced delivery variability and planning effort.
- Designed a trade promotion optimization engine for a global CPG firm that selected weekly price promotions under constraints like product-family conflicts, post-promotion dip, and cannibalization, maximizing the gross margin.
- Developed a skill-based resource scheduling model for a UK multimedia company to allocate show schedules and varied staff. Minimized costs while ensuring sequencing continuity based on past performance.
- Created a shipping order consolidation model that reduced terminal calls and improved vessel planning by clustering orders based on delivery direction and week, lowering total maritime handling costs.
Technologies: Python, Pandas, NumPy, SQL, PostgreSQL, Jupyter Notebook, Cbc, Gurobi, CPLEX, PuLP, Pyomo, Google OR-Tools, Linear Programming, Network Optimization, Vehicle Routing Problem (VRP), Constraint Programming, Git, Trade Promotion Optimization (TPO), Scheduling, Supply Chain Optimization, Amazon EC2, Algorithms, Heuristics, Operations Research, Mixed-integer Linear Programming, Communication, Mathematical Modeling, Decision Intelligence, SimPy, Vehicle Routing, Task Scheduling, FICO Xpress, Scientific Data Analysis, Back-end, Combinatorial Optimization, Analytics, Consumer Packaged Goods (CPG), Pricing Models, Data Science, Databricks
Assistant Business Analyst
2017 - 2019
Tata Consultancy Services
- Designed an MIP model to optimize truck loads by weight and volume using CPLEX, boosting average utilization above 91% for a CPG client’s 3rd-party logistics fleet across multiple regions.
- Developed a daily replenishment optimization model to align central distribution center shipments with monthly retailer targets and kept deviations under 5%.
- Conducted an in-depth analysis of the distributor management system and distributor–retailer flow data to identify bottlenecks, redesigned planning logic, and improved fill rates for retailers by 15% through stakeholder collaboration.
Technologies: Python, Pandas, NumPy, SQL, Git, CPLEX, Linear Programming, Operations Research, Computational Infrastructure for Operations Research (COIN-OR), Microsoft Excel, Microsoft PowerPoint, Supply Chain Optimization, Inventory Optimization, Network Optimization, Business Process Optimization, Logistics & Supply Chain, Stakeholder Engagement, Mixed-integer Linear Programming, Industrial Engineering, Communication, Mathematical Modeling, Decision Intelligence, Scientific Data Analysis, Combinatorial Optimization, Analytics, Consumer Packaged Goods (CPG), Data Science
Intern
2016 - 2016
Kansai Nerolac Paints
- Employed Pareto and 5-whys analysis to uncover root causes of production delays and designed Poka-yoke process sheets to eliminate manual errors.
- Used eVSM software to perform value stream mapping (VSM) and identify non-value-adding steps across the process, resulting in a streamlined workflow.
- Implemented the DMAIC process and reduced the cycle time of solid and metallic paint batches by 20% and 18% respectively.
Technologies: DMAIC, Microsoft Excel, Microsoft PowerPoint, Business Process Optimization, Productivity Improvement, Root-cause Analysis (RCA), Electronic Value Stream Mapping (eVSM), Value Stream Mapping, Communication, Scientific Data Analysis
Graduate Engineer Trainee
2014 - 2015
Gujarat Narmada Valley Fertilizers & Chemicals
- Analyzed vibration data from rotating equipment using Emonitor to enable predictive maintenance decisions and reduce unplanned downtime.
- Scheduled and allocated personnel in the fertilizer section based on equipment criticality and daily workload to optimize productivity and shift balance.
- Generated inspection reports and maintenance memos from field data, feeding into long-term reliability tracking and data-driven maintenance scheduling.
Technologies: SAP Quality Management (QM), Microsoft PowerPoint, Microsoft Excel, Statistical Analysis, Condition Monitoring, Predictive Maintenance, Productivity Improvement, Communication
Experience
Trade Promotion Optimization for a CPG Company
I developed an optimization solution for a global consumer goods company in the home and personal care space to generate the most profitable year-long trade promotion calendar.
The objective was to maximize incremental revenue by selecting the best weeks and discount levels to run promotions for various product groups across multiple retailers.
The model incorporated real-world constraints such as:
• Minimum and maximum weeks on promotion per product group.
• Mandatory cool-off periods after consecutive promotions.
• All SKUs in a product family must be promoted together.
• Cross-product effects due to cannibalization and complementarity.
• Post-promotion volume dips.
I built an MIP model in Python using Gurobi, which solved each instance optimally in under two minutes. The solution led to a 12% uplift in incremental revenue over the client’s manually curated plan for the UK market.
The objective was to maximize incremental revenue by selecting the best weeks and discount levels to run promotions for various product groups across multiple retailers.
The model incorporated real-world constraints such as:
• Minimum and maximum weeks on promotion per product group.
• Mandatory cool-off periods after consecutive promotions.
• All SKUs in a product family must be promoted together.
• Cross-product effects due to cannibalization and complementarity.
• Post-promotion volume dips.
I built an MIP model in Python using Gurobi, which solved each instance optimally in under two minutes. The solution led to a 12% uplift in incremental revenue over the client’s manually curated plan for the UK market.
Supply Chain Network Optimization for a Leading Cement Manufacturer
I worked with one of India’s largest cement manufacturers to optimize their production and distribution network across a multimodal (road and rail) supply chain.
The goal was to minimize total operational costs while ensuring demand fulfillment at over 2,000 dealer locations. The model considered production limits at plants and capacity constraints on each transport lane and allowed movement through distribution centers and regional warehouses.
I formulated the problem as a minimum-cost network flow model using MIP, where each facility type was a node and each transportation option an arc. I implemented the model in Python with Gurobi and solved it to optimality. The optimized network plan resulted in a 1.2% reduction in total logistics and production costs.
The goal was to minimize total operational costs while ensuring demand fulfillment at over 2,000 dealer locations. The model considered production limits at plants and capacity constraints on each transport lane and allowed movement through distribution centers and regional warehouses.
I formulated the problem as a minimum-cost network flow model using MIP, where each facility type was a node and each transportation option an arc. I implemented the model in Python with Gurobi and solved it to optimality. The optimized network plan resulted in a 1.2% reduction in total logistics and production costs.
Job Shop Scheduling for a Biopharmaceutical Client
A biopharmaceutical client needed an optimized production schedule for their multi-stage production process to meet tight delivery deadlines.
The objective was to minimize the average lateness and tardiness of all production orders, while also aiming to reduce the overall makespan. Each production order had a defined sequence of operations. Most processes had only one available machine, allowing no overlaps; only the fermentation stage had multiple parallel units.
I built a constraint programming model using Google OR-Tools and solved it using the CP-SAT solver to capture the complex sequencing and timing requirements. The model successfully generated near-on-time schedules while reducing the total makespan by 14%, improving planning efficiency and throughput in a tightly regulated production environment.
The objective was to minimize the average lateness and tardiness of all production orders, while also aiming to reduce the overall makespan. Each production order had a defined sequence of operations. Most processes had only one available machine, allowing no overlaps; only the fermentation stage had multiple parallel units.
I built a constraint programming model using Google OR-Tools and solved it using the CP-SAT solver to capture the complex sequencing and timing requirements. The model successfully generated near-on-time schedules while reducing the total makespan by 14%, improving planning efficiency and throughput in a tightly regulated production environment.
Education
2015 - 2017
Master's Degree in Manufacturing Management
Indian Institute of Management Mumbai - Mumbai, Maharashtra, India
2010 - 2014
Bachelor's Degree in Mechanical Engineering
Gujarat Technological University - Ahmedabad, Gujarat, India
Certifications
MAY 2020 - PRESENT
Discrete Optimization
Georgia Institute of Technology | via Coursera
JULY 2017 - PRESENT
Machine Learning
Stanford University | via Coursera
DECEMBER 2016 - PRESENT
Foundations of Business Strategy
University of Virginia | via Coursera
NOVEMBER 2016 - PRESENT
Supply Chain Operations
Rutgers, The State University of New Jersey | via Coursera
Skills
Libraries/APIs
Pandas, NumPy, Cbc, Ipopt, PySpark, Python API
Tools
Microsoft PowerPoint, Microsoft Excel, Git, CPLEX, Gurobi, Google OR-Tools, SAP Quality Management (QM), Amazon SageMaker, Hexaly
Languages
Python, SQL
Paradigms
Linear Programming, Constraint Programming, Mathematical Optimization, DMAIC
Frameworks
SimPy, Streamlit
Platforms
Jupyter Notebook, Databricks, Amazon EC2, AWS Lambda, Azure, Linux, Visual Studio Code (VS Code)
Industry Expertise
Trade Promotion Optimization (TPO)
Storage
PostgreSQL
Other
Mathematical Programming, Supply Chain Optimization, Network Optimization, Algorithms, Mathematical Modeling, Decision Intelligence, Mixed-integer Linear Programming, FICO Xpress, Scientific Data Analysis, Combinatorial Optimization, Pricing Models, Operations Research, Business Management, Industrial Engineering, Heuristics, Communication, Statistical Analysis, Business Process Optimization, Computational Infrastructure for Operations Research (COIN-OR), Inventory Optimization, PuLP, Pyomo, Vehicle Routing Problem (VRP), Scheduling, Cost Reduction & Optimization (Cost-down), XPRESS, Column Generation, Decomposition Methods, Quadratic Programming, Nonlinear Optimization, Genetic Algorithms, Monte Carlo Simulations, Taxation, Portfolio Analysis, Pricing Optimization, Vehicle Routing, Task Scheduling, AI Agents, Artificial Intelligence (AI), Back-end, Time Series Forecasting, Analytics, Consumer Packaged Goods (CPG), Data Science, Arena, Condition Monitoring, Predictive Maintenance, Productivity Improvement, Root-cause Analysis (RCA), Electronic Value Stream Mapping (eVSM), Value Stream Mapping, Logistics & Supply Chain, Stakeholder Engagement, Simulations, A/B Testing, eCommerce, eCommerce Fulfillment, Simulated Annealing, Prescriptive Analytics, Operations, Production Planning & Control (PPC), Optimization, Machine Learning, Supply Chain Operations, Warehouse Operations, Logistics, Distribution
Collaboration That Works
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
1
Share your needs
Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2
Choose your talent
Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3
Start your risk-free talent trial
Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.
Top talent is in high demand.
Start hiring