Anidipta P.

Key Skills

Software

CVAT

Data Annotation Tech

Labelbox

Label Studio

LabelImg

Mercor

SuperAnnotate

Roboflow

OpenCV AI Kit (OAK)

Top Subject Matter

Medical image analysis (ALL detection)

Computer vision research (high-resolution visual understanding)

Computer vision for vehicle landmark detection (road and lane)

Top Data Types

Image

Text

Document

Top Task Types

Fine-tuning

Segmentation

Bounding Box

Polygon

Classification

Object Detection

Question Answering

Text Generation

Red Teaming

Entity (NER) Classification

Text Summarization

Evaluation/Rating

Prompt + Response Writing (SFT)

Freelancer Overview

Machine Learning Engineer - Omdena. Brings 2+ years of professional experience across legal operations, contract review, compliance, and structured analysis. Education includes Bachelor of Technology, Heritage Institute of Technology, Kolkata (2027). AI-training focus includes data types such as Medical, DICOM, and Image and labeling workflows including Fine-tuning, Evaluation, and Rating.

Labeling Experience

AI Research Intern, HEVA AI Research Labs, India (Mar 2026 – May 2026)

ImageFine-tuning

Researched recurrent spatial memory architectures to enable efficient high-resolution visual understanding. The research supports downstream training/fine-tuning pipelines for computer vision models using spatial sequence memory mechanisms. This work focused on improving model representation quality for image-based recognition tasks. • Recurrent spatial memory architectures • Efficient visual understanding for high-resolution imagery • Model training research for visual recognition • Image-based computer vision experiments

2026 - 2026

Research Intern, NIT Rourkela (Jan 2026 – May 2026)

Fine-tuning

Designed a self-supervised Bayesian CNN workflow for ALL (acute lymphoblastic leukemia) detection using uncertainty-aware probabilistic modeling on limited medical datasets. The work implies preparing/curating labeled medical training sets for model learning and evaluation. The training process targeted improved generalization under data scarcity. • Self-supervised Bayesian CNN for ALL detection • Uncertainty-aware probabilistic modeling • Limited medical dataset curation for training • Under review publication activity tied to the training approach

2026 - 2026

Data Annotator

Geospatial Tiled ImageryBounding Box

Labelled 1050 MultiSpectral Images for agricultural domain crop handling

2025 - 2026

Machine Learning Engineer - Omdena

TextSegmentation

Engineered multiple applied machine learning solutions across diverse problem statements using LLM and classical ML techniques. Delivered an LLM-as-judge evaluation workflow using LangChain with structured JSON verdict generation for reliable output formatting. Built evaluation pipelines and forecasting systems, and implemented geospatial segmentation evaluation to improve downstream performance metrics. • Developed LLM-as-judge workflows and structured JSON verdict generation • Built eval pipelines with precision/recall metrics and expert calibration • Designed time-series forecasting with drift monitoring and error reduction • Implemented geospatial 3D segmentation evaluation to boost mIoU with low-confidence routing

2025 - 2025

AI/ML Intern, Vizzle (Jan 2025 – Mar 2025)

Built an end-to-end deep learning pipeline for real-time facial landmark detection to support virtual try-on functionality. Landmark detection requires supervised labeling/annotation of keypoints for training and evaluation. The project therefore involves creating or using labeled facial keypoint data to train the detection model. • Deep learning pipeline for real-time facial landmark detection • Supervised keypoint/landmark training and evaluation • Dataset labeling/annotation usage for model development • Support for AR virtual accessory try-on

2025 - 2025

Education

Work History