Dec 2023 – Dec 2024 · 1 yr

May 2023 – Oct 2023 · 5 mos

• Spearheaded the creation of H1's GenosAI Chatbot, enhancing user interaction and simplifying access to clinical trials data.
• Drove the transition of applied research to production, deploying technologies like Weaviate and ElasticSearch and fostering the use of retrieval augmented generation and code generation through large language models.
• Played an instrumental role in the launch of a robust MVP and the assembly of a high-performing team from ground zero, ensuring the achievement of key project milestones in sync with the company's strategic objectives.

Feb 2023 – Feb 2024 · 1 yr

• Led the design and development of an NLP agent, utilizing LLMs to build NLU, Summarization, and Retrieval Augmented Generation capabilities.
• Successfully integrated the NLP agent into the Compliance Bot product.

Nov 2022 – Feb 2023 · 3 mos · Bengaluru, Karnataka

Oct 2021 – Oct 2022 · 1 yr · Bengaluru, Karnataka, India · Hybrid

• Designed and developed RCAEngine for automated biometric image analysis, leading to significant time savings and reduced data retention period. Established a scalable CI/CD pipeline, designed self-service Tableau dashboards and completed the first project with a new India-based team.
• Designed and developed an Identity Analysis Framework using Spark on AWS EMR for identifying various user groups in real-time biometric recognition systems. Conducted research in Doddington Zoo Classification and Biometric Menagerie, providing insights to fine-tune recognition models and boost accuracy.

Dec 2020 – Oct 2021 · 10 mos

• Improved Document AI models and enabled distributed training, focusing on fine-tuning architectures like LayoutLM and Bert, Paddle OCR and inhouse models.
• Worked on NER using transformer architectures like LayoutLM, and improved existing training and evaluation pipeline.
• Developed a plug-and-play system for detecting, extracting, and classifying QR codes present in e-invoices using various computer vision algorithms. This system was eventually deployed across several business units for Amazon India.

May 2020 – Nov 2020 · 6 mos

• ● Converted the high-level business problem of getting insights from the logs generated by
• medical care-givers into various NLP problems.
• ● Worked on problems like NER, text classification, detection and localization of temporal
• event using BERT based models.
• ● Worked on various data augmentation techniques for handling class imbalance including
• methods like back translation.
• ● Built Web Crawlers and data cleansing pipeline for extracting caregiver data to boost the
• accuracy of various models.
• ● Designed, Developed and Deployed the entire NLP engine from scratch on AWS ECS.

Jun 2018 – Nov 2020 · 2 yrs 5 mos

• Worked on developing deep learning models using various architectures for precise agricultural produce detection, localization, and classification. Significantly enhanced image resolution, improved image preprocessing and model architecture, and applied advanced enhancement techniques for accurate defect classification. Conducted color science research to ensure generalization across different lighting conditions, and developed image processing techniques for precise dimension estimation of the produce. Also co-authored patents from research findings. Led the end-to-end development of the iFlow platform, automating processes, and saving 5 person-months per year. Additionally, designed and built the iProcess ML inference engine, enhancing its efficiency and performance through the use of various technologies.

Jul 2017 – Jun 2018 · 11 mos

• ● Researched various image segmentation and object detection techniques.
• ● Responsible for building the MVP from scratch for the seed funding round.

Apr 2018 – Jun 2018 · 2 mos

• ● Researched various pixel-based clustering and color distance measurement algorithms.
• ● Created a pipeline to extract dominant colors from catalog images with the goal to provide
• color based sorting of various product images.