Arshveer Kaur

Applied AI Specialist

Micro1

Aug 2025 - Jan 2026 (6 months)

• Designed LLM evaluation frameworks focused on reasoning quality, factual consistency, and instruction- following, combining automated metrics with structured human review to reduce failure modes. • Build prompt suites and test sets to iterate through error analysis and reduce recurring failure modes, and improve consistency. • Build annotation guidelines & scoring rubrics to standardize evaluations.

B

Machine Learning Scientist

Birla Institute of Technology & Science

Jun 2024 - Jun 2025 (1 year 1 month)

• Owned an end-to-end complex image time-series analytics pipeline: downloaded and managed 10+ TB data; built robust preprocessing using NumPy/Pandas to produce model-ready training datasets. • Designed and implemented a transformer architecture, leveraging masking-based learning tailored for com- plex, large-scale image sequences where baselines such as ViViT/VideoMAE underperformed. • The design improved performance by up to 30% while reducing GPU compute memory by 55%. • Optimized training and inference (profiling, batching, data loading, CUDA-aware optimizations), enabling scalable experiments and stable throughput on large datasets. • Deployed the model as a production service on an in-house Kubernetes platform and CUDA cluster (A100 G

B

Senior Research Fellow (ML)

Birla Institute of Technology and Science (BITS)

Jan 2018 - May 2024 (6 years 5 months)

• Built and evaluated deep learning systems (transformers, contrastive learning, multimodal fusion) for large- scale image time series data, achieving 25% error reduction on key tasks. • Designed and optimized end-to-end pipelines for data cleaning, preprocessing, feature construction, train- ing, and evaluation on 10+ TB datasets. • Improved efficiency of shipped research systems by revisiting modeling and pipeline trade-offs (compute vs. accuracy), leading to faster iteration cycles and lower training cost. • Delivered state-of-the-art multimodal fusion models and recommendation models for satellite analytics. • Led & mentored 10+ ML projects from conception to deployment, resulting in 10+ publications.

T

Assistant System Engineer

Tata Consultancy Services(TCS)

Aug 2016 - Dec 2017 (1 year 5 months)

○ Designed and implemented ETL (Extract, Transform, Load) pipelines to process and analyze large datasets, ensuring efficient data extraction, transformation, and integration across multiple systems. ○ Developed automated data workflows & optimized data extraction to improve processing efficiency. ○ Received Kudos Award (for top performer) during training

D

Back End Developer

Dynamic Verticals Softwares,

Jan 2014 - Jul 2014 (7 months)

○ Developed and optimized complex SQL queries to manage and process large-scale datasets for live, high-traffic web applications, ensuring reliability and performance. ○ Designed and maintained relational database schemas, enabling efficient data storage, retrieval, and integration to support dynamic website functionality. ○ Collaborated with front-end teams to provide seamless integration of database-driven features, ensuring a robust user experience for end users.

Arshveer Kaur

About

Detail

Timeline

Résumé