Devanshu Sharma

Physics Specialist (Llm Evaluation & Computational Modeling) @Mindrift

Aachen, DE
MOBILE NUMBERS
+91 *********19

Signup · Get unlimited contacts

WORK HISTORY

Jun 2025 — Present

Physics Specialist (Llm Evaluation & Computational Modeling) @Mindrift

View department →

Designed and validated complex computational frameworks to evaluate Large Language Model (LLM) performance in quantitative reasoning- Developed production-style Python implementations to stress-test model robustness and logical consistency- Analyzed output reliability using structured performance metrics and rigorous edge-case testing.

EDUCATION

N/A

RWTH Aachen University

Doctor of Philosophy - PhD, Physics

2015 — 2017

Shri Vardhaman Vidyalaya, Mehsana

High School, Science and Maths

2017 — 2022

Sardar Vallabhbhai National Institute of Technology, Surat

Integrated Master of Science - MSc, Physics

ABOUT DEVANSHU SHARMA

I am a Computational Physicist (PhD) specializing in stochastic modeling, large-scale numerical simulations, and uncertainty quantification. My expertise lies in building high-performance, modular pipelines to solve non-linear problems, whether they are found in the early universe or global energy markets. What I bring to the table:* Advanced Modeling: Expert in Stochastic Differential Equations (SDEs), Monte Carlo methods (VEGAS), and Bayesian inference.* High-Performance Computing: Experienced in accelerating scientific code by 30%+ using JAX, JIT compilation, and parallel computing.* Industry Application: Proven track record in applying Machine Learning and statistical models to electricity price forecasting and portfolio optimization under regime-dependent volatility.* Rigorous Validation: Deep experience in backtesting, sensitivity analysis, and evaluating model performance under extreme stress conditions. Technical Stack: Python (JAX, NumPy, SciPy, Pandas, Scikit-learn, Tensorflow), C, Mathematica, Linux/HPC, Git, SQL. Let’s connect if you’re looking for someone who can turn complex mathematical hypotheses into validated, high-performance code.

This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.