Open to opportunities

Abhinaash User

@abhinaashuser

Message

Entry-level data scientist building deep learning and scientific ML for forecasting, optimization, and vision.

Nepal

Message

What I'm looking for

I’m looking for a role where I can apply deep learning and scientific machine learning to real problems—building robust pipelines, optimizing models under constraints, and improving accuracy through careful validation.

I’m an entry-level data scientist who turns domain problems into working models—whether it’s portfolio optimization for NEPSE, physics-constrained learning, or computer vision. I developed a machine-learning-driven portfolio optimization framework using LSTM forecasting, a Ledoit–Wolf covariance shrinkage risk model, and Differential Evolution to satisfy cardinality and sector constraints.

I also build scientific ML systems with strong inductive bias: I implemented Physics-Informed Neural Networks (PINNs) to approximate heat equation solutions by enforcing PDE, boundary, and initial conditions, and benchmarked them against classical solvers. On the vision side, I trained a CNN-based facial keypoint recognition pipeline with OpenCV, and I explored image compression via Singular Value Decomposition using rank-based matrix approximation.

Experience

Work history, roles, and key accomplishments

Portfolio Optimization ML

Nepal Stock Exchange

Developed a machine-learning portfolio optimization framework for NEPSE, forecasting stock returns with LSTM using OHLCV data and walk-forward validation. Built a Ledoit–Wolf-based risk model with liquidity, transaction cost, and market impact adjustments, and solved a constrained portfolio optimization problem using Differential Evolution under cardinality and sector constraints.

Python PyTorch OHLCV Time Series Forecasting Walk Forward Validation Ledoit Wolf Covariance Shrinkage Differential Evolution Risk Modeling

Education

Degrees, certifications, and relevant coursework

Kathmandu University

Bachelor of Science, Computational Mathematics

Activities and societies: Relevant coursework: Algorithms, Operating Systems, Database Systems, Numerical Methods, Optimization, Probability & Statistics. Workshops/Schools: Annual Nepal AI School (NAAMI) Dec 2025–Jan 2026; Cryptography for the 21st Century (AESIM School) Aug 2025.

BSc in Computational Mathematics covering algorithms, operating systems, database systems, numerical methods, optimization, and probability & statistics.