Skip to content
View urmeo's full-sized avatar

Block or report urmeo

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
urmeo/README.md

Urme Bose

ML Engineer

Python · ML · Deep Learning · PyTorch

Portfolio · All repositories

12 public projects

Source → Code → Checks → Results

Projects

Project Focus Tech stack
CalmSense Wearable stress classification scikit-learn · PyTorch · React
NexusRAG Scientific paper retrieval LanceDB · Ollama · FastAPI
NovaVision Emotion conditioned images PyTorch · Diffusers · CLIP
retinonorm Synthetic receptive field fitting NumPy · SciPy
Multimodal-Multisensor Descriptive recordings pandas · SciPy · Jupyter
Sensor Device and source comparison Python · pandas
Psychometric Bilingual questionnaire scoring JavaScript · jsPDF
PulseShift Bike demand and weather scikit-learn · JavaScript
Antenna Patch antenna reports and checks CST · VBA · Python
shono Bengali transcription and scoring Python · jiwer
glassbox Interface scoring fixtures Python · HTML/CSS
portfolio Portfolio and accessibility HTML · CSS · JavaScript
Selected results
CalmSense saved binary subject holdout accuracy across five models
CalmSense

15 subjects · wearable stress
NexusRAG saved nDCG scores for two full retrieval test splits
NexusRAG

623 test queries · paper retrieval
NovaVision pilot label recovery with saved bootstrap intervals
NovaVision

49 pilot records · image recovery
CortexProbe saved synthetic position recovery across three noise levels
CortexProbe

5 synthetic fields · receptive field fitting
PulseShift saved temporal holdout AUROC for five demand models
PulseShift

8,204 holdout hours · bike demand
Psychometric item counts for five bilingual questionnaire instruments
Psychometric

88 items · bilingual scoring

Recorded results and authored scoring structures. Open each project for methods, sources and limits.

Pinned Loading

  1. CalmSense CalmSense Public

    Wearable biosignal classification, evaluation by subject, calibration and a static results dashboard.

    Python 5

  2. NexusRAG NexusRAG Public

    Local RAG for scientific papers: hybrid BM25 + dense retrieval, cited answers, SciFact/NFCorpus benchmark

    Python 4 1

  3. Multimodal-Multisensor Multimodal-Multisensor Public

    Eye, cardiac and questionnaire recordings from 10 adults across 3 sessions; descriptive analysis and source limits.

    Python 2

  4. NovaVision NovaVision Public

    Generate images from emotion labels; inspect a small CLIP emotion recovery pilot.

    Python