Looking for a job

Nakul Sinha

@nakulsinha

Machine Learning Engineer at Shipd, Datacurve, creating Kaggle-style challenges and expert solutions for frontier model training.

India

Message

At Shipd, Datacurve, I create and solve Kaggle-style machine learning challenges across vision, NLP, tabular, and time-series tasks, producing expert solutions used to train Datacurve’s frontier model. I also fine-tune vision and language models and contribute tasks to the DeepSWE benchmark.

At Jio Platforms Limited, I architected a distributed telecom analytics platform that cut turnaround from up to a day to under 10 minutes. I also designed anomaly detection pipelines and built a LangGraph framework that reduced incident investigation time from over four hours to under 15 minutes.

At Samsung SEED Lab, I contribute to ongoing research in machine learning, deep learning, computer vision, and NLP. My projects include Atlas, a vision-driven desktop agent, and Hermes, a decentralized messaging and file-transfer CLI.

Experience

Work history, roles, and key accomplishments

Current

Student Researcher

Current

Samsung SEED Lab

Oct 2026 - Present (1 month)

Contributing to ongoing research spanning machine learning, deep learning, computer vision and NLP.

Machine Learning Deep Learning Computer Vision NLP

Current

Machine Learning Engineer - Contract

Current

Shipd, Datacurve

Aug 2026 - Present (3 months)

Creating and solving 100+ Kaggle-style machine learning challenges across vision, NLP, tabular and time-series tasks, producing expert solutions used to train Datacurve's frontier model. Fine-tuning vision and language models and training gradient-boosted models, then combining them with grouped out-of-fold validation, greedy blending and metric-aware decoding.

Machine Learning Kaggle Computer Vision NLP Time Series ConvNext SigLIP DeBERTa ByT5 LoRAs Qwen3 8B LightGBM CatBoost Out Of Fold Validation Greedy Blending

Current

Research Director

Current

Association for Computing Machinery (ACM), VIT

May 2025 - Present (1 year 6 months)

Led Systems Reading Group sessions covering technical paper reviews, research problem formulation and methodology design. Organized and led large-scale technical events including z0d1ak CTF, Code2Create, Cryptic Hunt, Reverse Coding, and Neural Hack, overseeing problem design, evaluation pipelines, and system execution.

Research Event Management CTF Problem Design Evaluation Pipelines

Machine Learning Engineer - Intern

Jio Platforms Limited

May 2026 - Aug 2026 (3 months)

Architected a distributed telecom analytics platform ingesting 800+ CDR files per day, cutting end-to-end turnaround from up to a day to under 10 minutes. Designed anomaly detection pipelines and built an agentic LangGraph framework for automated feature selection and root-cause analysis.