Looking for a job

Onkar User

@onkar_thete

Message

Data Scientist building classification, anomaly detection, and NLP systems with RAG and agentic AI.

India

Message

What I'm looking for

I’m looking for work where I can ship NLP and ML prototypes into production—building classification and retrieval/agentic AI pipelines, collaborating with stakeholders, and turning messy requirements into deployed systems.

I build data-driven NLP and ML systems, from TF-IDF/LinearSVC document classification to radiology-focused NLI pipelines and hallucination detection workflows.

At Hexaware Technologies, I shipped a fintech document classification prototype in five days and later boosted accuracy to 94% by combining Tesseract OCR with LayoutLM for spatial-aware extraction.

At the University of Nottingham, I constructed OpenI-NLI from 3,412 clinical reports and deployed a FastAPI + SQLAlchemy hallucination detection pipeline with a React frontend for per-claim verdicts and confidence/evidence pointers.

Experience

Work history, roles, and key accomplishments

Data Science Intern

Infyntrek Systèmes

Apr 2026 - Jul 2026 (3 months)

Preprocessed and engineered features across structured and unstructured datasets to support ML model development. Executed end-to-end EDA pipelines to surface actionable patterns and statistical trends.

Python scikit learn Cross Validation EDA Feature Engineering Data Visualization

Research Assistant

University of Nottingham

May 2025 - Sep 2025 (4 months)

Constructed OpenI-NLI, a novel silver-labeled radiology NLI corpus from 3,412 real clinical reports. Domain-adapted BiomedBERT via LoRA achieving 96.6% accuracy and 0.978 F1 on entailment.

PyTorch Transformers BERT LoRAs NLP BM25 fastAPI SQLAlchemy React

Python Developer

Hexaware Technologies

Feb 2022 - Jul 2024 (2 years 5 months)

Built an end-to-end document classification system achieving 91% accuracy across 7 document types. Scaled classification accuracy to 94% by integrating Tesseract OCR and fine-tuning a LayoutLM model.

Python scikit learn TF IDF Tesseract LayoutLM Cross Validation