Vikram Naresh Email & Phone Number

Gurugram, Haryana, India

GenAI

Gurgaon, Haryana, India

• Engineered a Multi-task framework, a shared model encoder to consolidate similar models and generate common embeddings in a shared vector space. It efficiently replaces multiple models in production, which led to a 40% reduction (-100k USD/yr) in deployment costs.• Developed an AutoML framework, a scalable and reliable distributed system as a microservice architecture that facilitates large-scale automation of model training, validation, and deployment. It also includes features for… Show more • Engineered a Multi-task framework, a shared model encoder to consolidate similar models and generate common embeddings in a shared vector space. It efficiently replaces multiple models in production, which led to a 40% reduction (-100k USD/yr) in deployment costs.• Developed an AutoML framework, a scalable and reliable distributed system as a microservice architecture that facilitates large-scale automation of model training, validation, and deployment. It also includes features for indexing large datasets, tuning diverse models with custom parameters, and generating customizable metrics reports. It improved the dev-hours efficiency of data scientists and analysts multi-fold, and several products were automated end-to-end within this framework. Show less

Gurgaon, Haryana, India

• Built a Named Entity Recognition system to identify brands, individuals, and products in digital media space with precision and recall metrics at >85%• Built a Predictive Dialing framework to improve engageable efficiency (28% increase in agent-utilization rate) of callcenter agents by statistical modeling.• Involved in end-to-end deployment and maintenance of these services on Kubernetes clusters at scale

Gurgaon, India

• Built text-tagging models and validation scripts for existing Hierarchical Attention Network architectures for ease of classification of engageability between clients and their customers on social platforms, and built modified SpaCy/BERT classifiers for the same, using Google’s latest (at that time) bi-directional transformer architecture, which improved precision by 12%.• Worked on scalable dockerization of tagging tools.• Received a pre-placement offer for a full-time position based… Show more • Built text-tagging models and validation scripts for existing Hierarchical Attention Network architectures for ease of classification of engageability between clients and their customers on social platforms, and built modified SpaCy/BERT classifiers for the same, using Google’s latest (at that time) bi-directional transformer architecture, which improved precision by 12%.• Worked on scalable dockerization of tagging tools.• Received a pre-placement offer for a full-time position based on performance. Show less

Chennai Area, India

CV project on image classification

B.Tech, Computer Science & Engineering, 8.4 Cgpa

Activities and Societies: Maximus - Math club | Apekshaa - Social responsibility club | Vortex - CSE Dept.'s technical symposium, Head.

97.6%, Class 12 - Cbse

Majored in Computer Science 488/500 Overall 99/100 in Maths and Chemistry

10 Cgpa, Class 10 - Cbse

Studied Hindi, Sanskrit and French as additional languages.

Kg To Class 5