Bachelor Of Engineering

India

Principle AI Enginner

• I performed comprehensive dataset quality analysis for text classification, uncovering key biases and inconsistencies, which enhanced the accuracy of the DistilBERT-based classification model by 8%.
• I built an entity classification model using Gemma 9B, deployed it in production using VLLM, and achieved>95% accuracy across 20 entity categories.
•  I contributed to developing an RAG engine, designing key components such as prompt classification, relevant chunk identification, prompt template definition,

Machine Learning Lead

• I led the team in building a system for automated real estate listings using generative AI, computer vision, and prompt engineering, significantly improving listing quality and reducing manual effort.
• I led the efforts for developing a marketing content generation platform for LinkedIn using OpenAI APIs and Segmind API(for text to image), reducing content creation time by 50%.
• I optimized resume parsing using GPT-3.5 Turbo, reducing the cost per resume from 7 cents to 2 cents, leading to a 71% cost reduction.
• I built a zero-shot classification model for medical text, reducing false positives by 90%.

Senior Associate

Led a project on corporate event extraction from news feeds, enhancing financial event tracking and decision-making.

Machine Learning Lead

• I developed a signature and header detection system using Conditional Random Fields (CRF), reducing manual processing efforts with around 70% accuracy.
• Created an Image classification model with YOLO to categorize various documents like passports, death certificates, and medical forms.
• Designed and developed pipeline for an email conversation organization system, leveraging NLP techniques to classify and filter inclusive emails from vast datasets, reducing reviewer processing time by 30%.
• Implemented PII detection pipeline using Regex, NER, and NLP techniques, ensuring regulatory compliance, and making reviews faster by 5%.
• Developed a robust pipeline to extract 40+ attributes from the text (including SI units, chemical names, numbers, and ratios), improving writer productivity by 20%.

Product Technical Lead

• I led the design and development of a content moderation system using YOLO (object detection) and BERT-based text classification, enabling real-time content moderation.
• Built a call center analytics pipeline, leveraging DeepSpeech for audio transcription, speaker diarization, and text analytics for sentiment analysis and intent classification.
• Developed a sports highlights extraction system for live events, enhancing media analytics.
• Developed an AI-powered content mining platform, allowing clients to create custom data extraction templates with NLP techniques such as POS tagging, NER, and classification.