Pranav Sharma
I help teams make forecasting, pricing, and ML outputs more trustworthy, explainable, and usable in real decisions.
I build practical machine learning systems for decision support under uncertainty — across forecasting, pricing analytics, and risk-aware decisioning.
Most teams do not need more AI first. They need clearer answers to questions like:
- Can we trust this forecast?
- Where does the model fail?
- Is the output biased?
- Is the advanced model actually better?
- Can business users explain and challenge the recommendation?
Work With Me · View Labs · Read Blog · Resume · Contact
Labs: Live Demos + Repos
These are independent portfolio systems built to demonstrate how forecasting, pricing, and decision logic can be made more transparent, testable, and useful.
Forecasting Sandbox Lite
Benchmarking and model-selection workflows for time-series forecasting with transparent evaluation, bias-aware scoring, and reproducible comparisons.
Pricing Decision Lite
Scenario-based pricing analytics under demand uncertainty with explicit, rule-based decision support.
Decision Kernel Lite
Structured decision analysis using Expected Loss, Minimax Regret, and CVaR for comparing actions under uncertainty.
Work With Me
I am a strong fit for teams that already have forecasting, pricing, or ML workflows, but need stronger evaluation, uncertainty handling, and decision logic.
Typical areas:
- forecasting system audits
- model evaluation and benchmarking
- probabilistic forecasting
- pricing analytics
- decision intelligence prototypes
- MLOps / AI Operations reviews
About
I am a Senior AI Engineer based in Heilbronn, Germany.
My work focuses on operational ML systems for forecasting, pricing analytics, decision intelligence, and AI Operations. I care less about isolated model demos and more about whether the system can be evaluated, explained, deployed, and trusted in real workflows.
Compliance Note
This site contains independent portfolio content only. Views are my own and do not represent any employer. No proprietary or confidential employer data, code, methods, or client information is included. All public demos use public or synthetic data.