Looking for a job

Dhruv Ramani

@dhruvramani

Message

Machine learning and computer vision developer building end-to-end NLP and CV systems with real-world accuracy.

India

Message

What I'm looking for

I’m looking for hands-on work where I can build and deploy end-to-end ML systems—especially computer vision and NLP—iterate fast on model quality, and ship real-time features with measurable impact.

I’m a B.Tech Computer Science & AI student with hands-on experience building and deploying end-to-end ML pipelines, computer vision systems, and NLP applications. I developed a Random Forest + XGBoost ensemble achieving 91% accuracy on NASA TESS exoplanet data, and built a real-time pose estimation coaching system using MediaPipe and TensorFlow.

I enjoy turning raw data into dependable predictions—whether that’s classifying exoplanet transit signatures from NASA MAST lightcurves or validating exercise form with geometric inference and a Finite State Machine. I’m motivated by practical, end-to-end problem solving and clear performance improvements from feature engineering to model calibration and deployment.

Experience

Work history, roles, and key accomplishments

Current

Exoplanet Detection Pipeline

Current

Newton School of Technology

Feb 2025 - Present (1 year 4 months)

Built an end-to-end exoplanet transit classification pipeline using NASA TESS lightcurves from the MAST archive, engineering 16 astrophysical features after stellar detrending and TLS period search. Trained a Random Forest + XGBoost voting ensemble that achieved 91% classification accuracy and packaged predictions with confidence scores and uncertainty quantification.

Python scikit learn XGBoost PyTorch Feature Engineering Random Forest Transit Least Squares (TLS)Wotan Biweight Filter Uncertainty Quantification

Current

Mavis Real-Time Pose Coach

Current

Newton School of Technology

Nov 2025 - Present (7 months)

Developed a real-time pose-estimation fitness coach that validates exercise form from a live webcam feed using MediaPipe Pose and TensorFlow with a Finite State Machine (FSM). Implemented geometric joint-angle inference to detect asymmetry, elbow drift, and tempo constraints for bicep curls and shoulder press.

Python OpenCV MediaPipe Pose TensorFlow Finite State Machines Pose Estimation Geometric Inference Joint Angle Analysis

Current

Research Paper Summarization

Current

Newton School of Technology

Mar 2025 - Present (1 year 3 months)

Created an NLP pipeline to retrieve, analyze, and summarize academic research papers across AI, Physics, Mathematics, and Quantitative Biology. Implemented TF-IDF + cosine similarity for query-based retrieval, LDA for topic modeling, and TextRank for extractive summarization.

Python SpaCy TF IDF Cosine Similarity Latent Dirichlet Allocation (LDA)TextRank Topic modeling

Education

Degrees, certifications, and relevant coursework

Newton School of Technology (Rishihood University)

B.Tech, Computer Science and Artificial Intelligence

2024 -

Activities and societies: Relevant coursework: Neural Networks and Deep Learning, Generative AI, Linear Algebra, Statistics, Data Analytics and Visualization. Projects: Exoplanet detection ML pipeline (NASA TESS data, RF+XGBoost ~91% accuracy) and real-time pose estimation fitness coach using MediaPipe + TensorFlow.

B.Tech in Computer Science and Artificial Intelligence (currently in the 2nd year). Working on coursework and projects in neural networks, generative AI, and data analytics.