Dhruv Mohta

Software Engineer

Mumbai, India / Leuven, Belgium 0+ yrs exp 86 · Excellent

About

Dual-degree engineering student (B.Sc. Electromechanical Eng., KU Leuven; B.S. Data Science and Programming, IIT Madras, expected Aug 2027), currently an Algorithms Developer at Atlas Copco building real-time control software in Python and C++. Comfortable across systems programming, backend services, and applied machine learning, with strong fundamentals in data structures, algorithms, OOP, and databases.

Skills & Expertise (27)

Python Intermediate

7.5/10

Years Exp

Pandas Intermediate

7.5/10

Years Exp

NumPy Intermediate

7.5/10

Years Exp

scikit-learn Intermediate

7.5/10

Years Exp

Transformers Intermediate

7.5/10

Years Exp

PyTorch Intermediate

7.5/10

Years Exp

RAG Intermediate

6.6/10

Years Exp

OOP Intermediate

6.6/10

Years Exp

Data Structures and Algorithms Intermediate

6.6/10

Years Exp

C++ Intermediate

6.6/10

Years Exp

FAISS Intermediate

6.6/10

Years Exp

Postgresql Intermediate

5.5/10

Years Exp

SQLite Intermediate

5.5/10

Years Exp

SQL Intermediate

5.5/10

Years Exp

REST APIs Intermediate

5.5/10

Years Exp

LINUX Intermediate

5.5/10

Years Exp

Git Intermediate

5.5/10

Years Exp

Docker Intermediate

5.5/10

Years Exp

Azure Intermediate

5.5/10

Years Exp

OpenCV Intermediate

5.5/10

Years Exp

Node.js Beginner

4.4/10

Years Exp

React Beginner

4.4/10

Years Exp

Express Beginner

4.4/10

Years Exp

Typescript Beginner

4.4/10

Years Exp

Java Beginner

4.4/10

Years Exp

FastAPI Beginner

4.4/10

Years Exp

Flask Beginner

4.4/10

Years Exp

Work Experience

Algorithms Developer

Atlas Copco Airpower nv, Cyberphysical Department

Oct 2025 - May 2026

Built real-time adaptive control software for industrial compressors, combining PID auto-tuning with a Proximal Policy Optimisation (PPO) reinforcement-learning layer for set-point tracking. Implemented an online-learning loop (2 sensor inputs, 1 actuator output) that retunes continuously in production without manual recalibration. Refactored the PPO policy from a deep MLP to a linear mapping, removing overfitting on a low-dimensional control task and cutting inference compute by over 99%, which made on-device deployment feasible. Porting the Python training and inference pipeline to C++ with LibTorch to integrate with the proprietary control stack: low-latency inference and deterministic execution on embedded hardware.